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 | 4,298,772,117 | I_kwDOQb6kR88AAAABADoOlQ | 69,470 | https://github.com/openclaw/openclaw/issues/69470 | https://api.github.com/repos/openclaw/openclaw/issues/69470 | NO_REPLY silent reply suppression fails when model prefaces with reasoning or emits doubled token | ## Bug
When an agent is configured to reply with the `NO_REPLY` sentinel (for example during spam/ad handling in Discord), the agent's model can emit reasoning as plain text content instead of (or in addition to) structured thinking events. The resulting message is posted to the messaging channel verbatim.
Observed D... | closed | not_planned | true | 2 | [] | [] | 2026-04-20T21:19:02Z | 2026-04-29T04:35:23Z | 2026-04-26T16:03:13Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | pdd-cli | 262,266,283 | U_kgDOD6Hdqw | User | false |
openclaw/openclaw | 4,298,776,114 | I_kwDOQb6kR88AAAABADoeMg | 69,471 | https://github.com/openclaw/openclaw/issues/69471 | https://api.github.com/repos/openclaw/openclaw/issues/69471 | Haiku requests show 0% cache hit rate — is prompt caching applied for anthropic/claude-haiku-4-5? | ### Summary
Our Anthropic usage reports show **0% cache hit rate on Haiku 4.5 requests** routed through the OpenClaw gateway, while Sonnet 4.6 consistently hits 44–71% on the same days. Asking whether this is expected behavior or a configuration issue.
### Environment
- OpenClaw: `2026.4.11 (769908e)` (Homebrew cask, ... | closed | completed | true | 2 | [] | [] | 2026-04-20T21:19:54Z | 2026-04-28T04:49:10Z | 2026-04-25T19:09:59Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | pl-cpu | 263,160,351 | U_kgDOD6-CHw | User | false |
openclaw/openclaw | 4,068,385,205 | I_kwDOQb6kR87yfqG1 | 44,550 | https://github.com/openclaw/openclaw/issues/44550 | https://api.github.com/repos/openclaw/openclaw/issues/44550 | [Bug]: agents.defaults.models.*.params.num_ctx ignored for Ollama — model discovery always overrides with GGUF context_length | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
OpenClaw ignores user-configured num_ctx in openclaw.json when calling Ollama; on every startup it queries /api/show, reads the GGUF metadata field qwen2.context_length: 32768, and passes that value as num_ctx in every /api/chat request, ma... | closed | not_planned | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-03-13T02:27:05Z | 2026-04-28T04:40:40Z | 2026-04-26T03:03:42Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | armi0024 | 5,780,255 | MDQ6VXNlcjU3ODAyNTU= | User | false |
openclaw/openclaw | 4,298,933,566 | I_kwDOQb6kR88AAAABADyFPg | 69,475 | https://github.com/openclaw/openclaw/issues/69475 | https://api.github.com/repos/openclaw/openclaw/issues/69475 | Skill SKILL.md description + size fixes (per Anthropic skill spec) | ## Summary
5 built-in OpenClaw skills fail the Anthropic Skill spec's rules for `description` field content and SKILL.md size. Filing as a single consolidated issue covering 4 HIGH (description field missing WHAT+WHEN) + 1 MEDIUM (size-cap exceeded).
All findings are from an independent audit applying a 10-rule skill... | open | null | false | 2 | [] | [] | 2026-04-20T21:55:59Z | 2026-05-02T16:01:51Z | null | NONE | null | 20260502T180029Z | 2026-05-02T18:00:29Z | pescaohq-web | 257,331,994 | U_kgDOD1aTGg | User | false |
openclaw/openclaw | 4,068,560,339 | I_kwDOQb6kR87ygU3T | 44,595 | https://github.com/openclaw/openclaw/issues/44595 | https://api.github.com/repos/openclaw/openclaw/issues/44595 | [Bug]: Telegram polling stall recovery fails — watchdog tracks call initiation, not success | ## Bug
Telegram polling stall detection correctly fires after ~90s of silence, but the **recovery fails** because the watchdog timestamp is updated on `getUpdates` call **initiation**, not **successful completion**. grammY's internal retry mechanism (`maxRetryTime: 3600000`) makes failed `getUpdates` calls at interval... | closed | completed | true | 1 | [] | [] | 2026-03-13T03:19:18Z | 2026-04-23T04:22:21Z | 2026-04-20T22:19:38Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | talison | 6,170 | MDQ6VXNlcjYxNzA= | User | false |
openclaw/openclaw | 4,095,119,325 | I_kwDOQb6kR870Fo_d | 49,822 | https://github.com/openclaw/openclaw/issues/49822 | https://api.github.com/repos/openclaw/openclaw/issues/49822 | [Bug]: getUpdates conflict (409) persists even with fresh bot token and health monitor disabled | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
OpenClaw's Telegram provider produces continuous `409 Conflict: terminated by other getUpdates request` errors even with a brand new Telegram bot token and health monitor completely disabled. This indicates an internal polling loop leak, no... | closed | completed | true | 7 | [
"bug",
"bug:behavior"
] | [] | 2026-03-18T13:24:25Z | 2026-04-23T04:22:20Z | 2026-04-20T22:19:41Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | vinceyap88 | 12,798,134 | MDQ6VXNlcjEyNzk4MTM0 | User | false |
openclaw/openclaw | 4,080,947,820 | I_kwDOQb6kR87zPlJs | 48,029 | https://github.com/openclaw/openclaw/issues/48029 | https://api.github.com/repos/openclaw/openclaw/issues/48029 | Telegram long-polling stalls permanently after NAT drops idle TCP connections — undici connection pool not recycled on restart | ## Description
When running openclaw gateway behind NAT/cloud firewall, Telegram long-polling periodically stalls and **cannot recover** without a full process restart. The gateway correctly detects the stall ("no getUpdates response for 90s") and attempts to restart polling, but the restart reuses the same undici HTT... | closed | completed | true | 2 | [] | [] | 2026-03-16T08:19:51Z | 2026-04-23T04:22:20Z | 2026-04-20T22:19:43Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | jackiedepp | 31,266,295 | MDQ6VXNlcjMxMjY2Mjk1 | User | false |
openclaw/openclaw | 4,141,297,905 | I_kwDOQb6kR8721zDx | 54,992 | https://github.com/openclaw/openclaw/issues/54992 | https://api.github.com/repos/openclaw/openclaw/issues/54992 | Telegram long polling can leave getUpdates stuck far longer than expected | ## Summary
Telegram long polling can get stuck in a way that is only partially visible from current logs.
In a real deployment, Telegram health probes remained healthy, but OpenClaw repeatedly logged polling stalls and forced restarts. After instrumenting the polling session, the observed pattern was:
- an in-flight... | closed | completed | true | 1 | [] | [] | 2026-03-26T07:12:55Z | 2026-04-23T04:22:19Z | 2026-04-20T22:19:45Z | CONTRIBUTOR | null | 20260423T060020Z | 2026-04-23T06:00:20Z | sinogello | 4,157,921 | MDQ6VXNlcjQxNTc5MjE= | User | false |
openclaw/openclaw | 4,078,666,120 | I_kwDOQb6kR87zG4GI | 47,458 | https://github.com/openclaw/openclaw/issues/47458 | https://api.github.com/repos/openclaw/openclaw/issues/47458 | bug(telegram): Polling stall loop — getUpdates hangs, restart never recovers (since 2026.3.7) | ## Summary
Telegram long-polling enters an unrecoverable stall loop after upgrading to `2026.3.7`. The poller detects no `getUpdates` for ~15 minutes, attempts restart, times out after 15s, force-restarts, and the cycle repeats indefinitely. This also causes `sendMessage` failures (network request errors).
## Environ... | closed | completed | true | 6 | [] | [] | 2026-03-15T16:10:15Z | 2026-04-23T04:22:18Z | 2026-04-20T22:19:47Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | nishantkabra77 | 194,645,679 | U_kgDOC5oOrw | User | false |
openclaw/openclaw | 4,299,040,567 | I_kwDOQb6kR88AAAABAD4nNw | 69,480 | https://github.com/openclaw/openclaw/issues/69480 | https://api.github.com/repos/openclaw/openclaw/issues/69480 | [Bug]: Workspace context files silently truncated at ~12,000 characters (AGENTS.md / gateway agents) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Workspace context files injected into the agent prompt (specifically `AGENTS.md`) are silently truncated at approximately **12,000 characters** for agents running via the OpenClaw Gateway (`openclaw_gateway` ag... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-04-20T22:24:38Z | 2026-04-24T04:26:04Z | 2026-04-21T18:53:01Z | NONE | null | 20260424T060021Z | 2026-04-24T06:00:21Z | edmod209 | 24,400,367 | MDQ6VXNlcjI0NDAwMzY3 | User | false |
openclaw/openclaw | 4,299,112,064 | I_kwDOQb6kR88AAAABAD8-gA | 69,482 | https://github.com/openclaw/openclaw/issues/69482 | https://api.github.com/repos/openclaw/openclaw/issues/69482 | [Bug]: Telegram /approve allow-always writes 'source' field that openclaw approvals set --gateway rejects — allowlist updates silently fail to persist to gateway | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Telegram "Allow always" approvals write an allowlist entry with a `source: "allow-always"` string field. The file write to ~/.openclaw/exec-approvals.json succeeds, but `openclaw approvals set --gateway --file ... | closed | not_planned | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-04-20T22:46:10Z | 2026-04-29T04:42:43Z | 2026-04-26T06:58:40Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | reidperyam | 4,805,874 | MDQ6VXNlcjQ4MDU4NzQ= | User | false |
openclaw/openclaw | 4,298,999,121 | I_kwDOQb6kR88AAAABAD2FUQ | 69,478 | https://github.com/openclaw/openclaw/issues/69478 | https://api.github.com/repos/openclaw/openclaw/issues/69478 | [Bug]: enqueueSystemEvent not deduplicated by runId/contextKey — agents cascade duplicate exec approval prompts under new IDs, locking ecosystem | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
## Summary
Under load, `enqueueSystemEvent` does not deduplicate queued exec approval requests by `runId` or `contextKey`. When a heartbeat run times out and the gateway fails over, the replacement attempt re-queues the **s... | open | null | false | 2 | [
"bug",
"regression"
] | [] | 2026-04-20T22:13:23Z | 2026-05-02T14:36:37Z | null | NONE | null | 20260502T150030Z | 2026-05-02T15:00:30Z | reidperyam | 4,805,874 | MDQ6VXNlcjQ4MDU4NzQ= | User | false |
openclaw/openclaw | 4,299,172,362 | I_kwDOQb6kR88AAAABAEAqCg | 69,484 | https://github.com/openclaw/openclaw/issues/69484 | https://api.github.com/repos/openclaw/openclaw/issues/69484 | Google provider: image/vision fails with 400 via OpenAI-compatible endpoint | ## Bug
When using `google/gemini-2.5-flash` (or `gemini-2.5-pro`) as `imageModel`, the image tool returns **400 (no body)**.
## Root Cause
OpenClaw sends image requests through Google's **OpenAI-compatible endpoint** (`/v1beta/openai/chat/completions`), which doesn't properly handle image inputs.
## Evidence
### O... | closed | completed | true | 2 | [] | [] | 2026-04-20T22:59:53Z | 2026-04-26T04:29:25Z | 2026-04-24T01:06:16Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | igorviapaineis | 221,347,750 | U_kgDODTF_pg | User | false |
openclaw/openclaw | 4,100,671,856 | I_kwDOQb6kR870a0lw | 50,367 | https://github.com/openclaw/openclaw/issues/50367 | https://api.github.com/repos/openclaw/openclaw/issues/50367 | polling-session: #confirmPersistedOffset() can hang indefinitely without client-side timeout | ## Description
The `#confirmPersistedOffset()` method in `src/telegram/polling-session.ts` (lines 171-181) calls `getUpdates({ offset: lastUpdateId + 1, limit: 1, timeout: 0 })` without `AbortSignal.timeout()` on the client side.
```typescript
async #confirmPersistedOffset(bot: TelegramBot): Promise<void> {
// ...
... | closed | completed | true | 2 | [] | [] | 2026-03-19T10:10:52Z | 2026-04-23T04:22:17Z | 2026-04-20T23:04:17Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | boticlaw | 262,230,840 | U_kgDOD6FTOA | User | false |
openclaw/openclaw | 4,299,191,654 | I_kwDOQb6kR88AAAABAEB1Zg | 69,487 | https://github.com/openclaw/openclaw/issues/69487 | https://api.github.com/repos/openclaw/openclaw/issues/69487 | [Bug]: First dreaming-narrative subagent run after gateway idle exceeds NARRATIVE_TIMEOUT_MS due to plugin-host cold-start | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
On a gateway that has been idle long enough for the subagent runtime's plugin host to cool, the first `memory-core` dreaming-narrative `subagent.run` of the morning sweep routinely exceeds the hard-coded 60 s `... | closed | completed | true | 1 | [] | [] | 2026-04-20T23:04:32Z | 2026-04-28T04:49:09Z | 2026-04-25T19:10:05Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | omerekin | 36,278,311 | MDQ6VXNlcjM2Mjc4MzEx | User | false |
openclaw/openclaw | 4,155,274,728 | I_kwDOQb6kR873rHXo | 55,987 | https://github.com/openclaw/openclaw/issues/55987 | https://api.github.com/repos/openclaw/openclaw/issues/55987 | [Bug]: Browser CLI and agent tool browser.request return UNAVAILABLE while browser control HTTP API and CDP work fine | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
All openclaw browser CLI commands and the agent browser tool fail with GatewayClientRequestError: UNAVAILABLE even though the browser control HTTP API on port 18791 and CDP on port 18800 are both running and reachable; the g... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-03-27T19:09:22Z | 2026-04-28T05:22:42Z | 2026-04-25T11:22:32Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | jennyftt1 | 267,773,901 | U_kgDOD_XnzQ | User | false |
openclaw/openclaw | 4,059,586,325 | I_kwDOQb6kR87x-F8V | 43,381 | https://github.com/openclaw/openclaw/issues/43381 | https://api.github.com/repos/openclaw/openclaw/issues/43381 | [Bug]: Gateway shows running, health check says disconnected (1006) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When starting the gateway, the gateway signals that it is running, but openclaw still acts like the gateway is disconnected.
I normally run openclaw though a wrapper, but ran a test that used:
- separate OPENCLAW_HOME
- separate config
- s... | closed | completed | true | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-03-11T17:25:24Z | 2026-04-23T04:22:16Z | 2026-04-20T23:29:16Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | ryanrlindsay | 250,583,498 | U_kgDODu-Zyg | User | false |
openclaw/openclaw | 4,299,233,751 | I_kwDOQb6kR88AAAABAEEZ1w | 69,492 | https://github.com/openclaw/openclaw/issues/69492 | https://api.github.com/repos/openclaw/openclaw/issues/69492 | system-event-shape consumer-path leakage; propose per-event audience classification | ## Summary
Under umbrella #69208 Track B ("Delivery-mirror and consumer-path leakage"), the 2026-04-20 cleanup comment names #39469 as the Track B canonical and scopes the supporting issues (#38061, #33263, #5964) to *assistant-message-shape* artifacts. PR #69217 is the active Track B fix for that slice — it adds `isT... | open | null | false | 4 | [] | [] | 2026-04-20T23:13:52Z | 2026-05-02T16:41:09Z | null | CONTRIBUTOR | null | 20260502T180029Z | 2026-05-02T18:00:29Z | lukeboyett | 46,942,646 | MDQ6VXNlcjQ2OTQyNjQ2 | User | false |
openclaw/openclaw | 4,299,337,205 | I_kwDOQb6kR88AAAABAEKt9Q | 69,497 | https://github.com/openclaw/openclaw/issues/69497 | https://api.github.com/repos/openclaw/openclaw/issues/69497 | [Bug]: ◇ Select channel (QuickStart) │ Skip for now TypeError: Cannot read properties of undefined (reading 'trim') | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
I the openclaw onboard menu, I encountered this error after clicking skip for now on the Select channel dialogue
◇ Select channel (QuickStart)
│ Skip for now
TypeError: Cannot read properties of undefined (reading 'trim')
... | closed | completed | true | 1 | [
"bug",
"close:already-fixed",
"dedupe:child",
"regression"
] | [
"vincentkoc"
] | 2026-04-20T23:43:20Z | 2026-04-28T05:29:05Z | 2026-04-25T09:37:16Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | MdungeGeneral | 32,886,387 | MDQ6VXNlcjMyODg2Mzg3 | User | false |
openclaw/openclaw | 4,190,136,276 | I_kwDOQb6kR875wGfU | 59,265 | https://github.com/openclaw/openclaw/issues/59265 | https://api.github.com/repos/openclaw/openclaw/issues/59265 | [Bug]: v2026.4.1 Agents are working in secret; no actions are visible in the chat. | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Agents and outputs are no longer visible in chat. This happened immediately after updating to v2026.4.1. No chats are currently being displayed in the OPC chat history. Additionally, Telegram messages are only... | closed | completed | true | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-04-01T21:59:49Z | 2026-04-28T05:14:08Z | 2026-04-25T13:02:08Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | LukaszRT | 103,423,432 | U_kgDOBiodyA | User | false |
openclaw/openclaw | 4,299,357,286 | I_kwDOQb6kR88AAAABAEL8Zg | 69,498 | https://github.com/openclaw/openclaw/issues/69498 | https://api.github.com/repos/openclaw/openclaw/issues/69498 | [Bug]: openclaw plugins install nests a full copy of openclaw via peer-dep auto-install — 800+ MB bloat and ~15s startup tax per command | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
`openclaw plugins install` runs `npm install` in the plugin directory, and npm 7+'s auto-install-peers behavior installs `openclaw` itself as a nested dependency of the plugin, dragging in openclaw's entire transitive tree (... | closed | completed | true | 2 | [] | [
"vincentkoc"
] | 2026-04-20T23:49:02Z | 2026-04-28T04:39:11Z | 2026-04-26T03:21:14Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Adam-Researchh | 261,827,313 | U_kgDOD5sq8Q | User | false |
openclaw/openclaw | 4,169,719,612 | I_kwDOQb6kR874iN88 | 57,660 | https://github.com/openclaw/openclaw/issues/57660 | https://api.github.com/repos/openclaw/openclaw/issues/57660 | [Bug]: Telegram polling stall detector fires too aggressively (110s), causes message delivery failures | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Telegram polling stall detector fires at 110s, restarting the gateway mid-message-generation and causing message delivery failures
### Steps to reproduce
1. Send a Telegram message to an OpenClaw bot that req... | closed | completed | true | 0 | [
"bug",
"bug:behavior"
] | [] | 2026-03-30T10:56:44Z | 2026-04-23T04:22:14Z | 2026-04-21T00:03:06Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | mclee1975 | 267,909,583 | U_kgDOD_f5zw | User | false |
openclaw/openclaw | 4,299,474,055 | I_kwDOQb6kR88AAAABAETEhw | 69,503 | https://github.com/openclaw/openclaw/issues/69503 | https://api.github.com/repos/openclaw/openclaw/issues/69503 | [Feature]: xAI TTS/STT Speech Provider | ### Summary
Add native xAI TTS/STT speech provider support for cost-effective voice features
### Problem to solve
OpenClaw currently lacks native support for xAI's TTS/STT APIs, which offer significantly lower costs (~$2-6/M tokens) and high-quality voices. Users who want cost-effective voice features must implement... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-04-21T00:20:47Z | 2026-04-25T04:15:49Z | 2026-04-23T03:30:24Z | NONE | null | 20260425T060021Z | 2026-04-25T06:00:21Z | Continuum360 | 9,889,832 | MDQ6VXNlcjk4ODk4MzI= | User | false |
openclaw/openclaw | 4,299,453,972 | I_kwDOQb6kR88AAAABAER2FA | 69,502 | https://github.com/openclaw/openclaw/issues/69502 | https://api.github.com/repos/openclaw/openclaw/issues/69502 | Enhancement: MCP client support in agent skills (enable imap-smtp-email and other MCP servers as backends) | ## Background
While analyzing options for improving the an `email` skill, I investigated several IMAP/SMTP MCP servers:
- [gabigabogabu/email-mcp-server](https://github.com/gabigabogabu/email-mcp-server) (Node.js/TypeScript) -- Tools: `send_email`, `search_emails`, `list_folders`; Resources: `mailto:/inbox`, `mailto:... | closed | completed | true | 0 | [] | [] | 2026-04-21T00:15:47Z | 2026-04-23T04:22:13Z | 2026-04-21T00:18:06Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | davidsulpy | 937,796 | MDQ6VXNlcjkzNzc5Ng== | User | false |
openclaw/openclaw | 4,299,501,384 | I_kwDOQb6kR88AAAABAEUvSA | 69,505 | https://github.com/openclaw/openclaw/issues/69505 | https://api.github.com/repos/openclaw/openclaw/issues/69505 | Feature Request: assistantPrefix on agentTurn payload (and agent defaults) to guarantee reply starts with fixed text | ## Summary
Add an optional `assistantPrefix` (or equivalent) string field to `CronAgentTurnPayloadFields` and to the agent-defaults config, so operators can force the assistant's reply to start with a specific string. This closes the door on preamble/narration at the protocol level, independently of how well-aligned t... | open | null | false | 1 | [] | [] | 2026-04-21T00:29:14Z | 2026-05-02T16:41:10Z | null | NONE | null | 20260502T180029Z | 2026-05-02T18:00:29Z | zote | 35,069 | MDQ6VXNlcjM1MDY5 | User | false |
openclaw/openclaw | 4,299,449,533 | I_kwDOQb6kR88AAAABAERkvQ | 69,501 | https://github.com/openclaw/openclaw/issues/69501 | https://api.github.com/repos/openclaw/openclaw/issues/69501 | memory subcommands (status / index / search / promote) hang indefinitely with no output on Linux containers | ## Summary
`openclaw memory <subcommand>` (status / index / search / promote) hangs forever on Linux containers, producing **zero bytes** of stdout/stderr. The CLI never writes anything, never errors, never exits. `openclaw --help` and `openclaw memory --help` work fine — only the actual subcommand handlers hang.
Rep... | closed | completed | true | 2 | [] | [] | 2026-04-21T00:14:42Z | 2026-04-28T04:49:08Z | 2026-04-25T19:10:12Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | rschumann | 360,788 | MDQ6VXNlcjM2MDc4OA== | User | false |
openclaw/openclaw | 4,299,530,936 | I_kwDOQb6kR88AAAABAEWiuA | 69,507 | https://github.com/openclaw/openclaw/issues/69507 | https://api.github.com/repos/openclaw/openclaw/issues/69507 | [Bug]: OpenRouter model redirects not handled properly in OpenClaw model resolution | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
When using OpenRouter model aliases (e.g., openrouter/deepseek/deepseek-v3.2), OpenClaw fails to handle the HTTP redirects properly, resulting in model resolution failures.
### Steps to reproduce
1. Specify O... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-21T00:39:06Z | 2026-04-29T04:42:43Z | 2026-04-26T06:58:44Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | ionblue88 | 277,917,499 | U_kgDOEJCvOw | User | false |
openclaw/openclaw | 4,299,543,623 | I_kwDOQb6kR88AAAABAEXURw | 69,510 | https://github.com/openclaw/openclaw/issues/69510 | https://api.github.com/repos/openclaw/openclaw/issues/69510 | [Bug]: Dreaming memory files incorrectly appear as agents in Web UI agent switcher | ## Bug Description
Markdown files under `memory/dreaming/` (e.g. `deep/2026-04-21.md`, `rem/2026-04-21.md`, `light/2026-04-21.md`) are incorrectly listed as selectable agents in the Web UI top-left agent switcher dropdown.
These files are auto-generated dreaming/memory log files and should not appear in the agent lis... | closed | not_planned | true | 1 | [] | [] | 2026-04-21T00:43:21Z | 2026-04-28T04:49:07Z | 2026-04-25T19:10:18Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | aoaeenft | 275,716,279 | U_kgDOEG8Ytw | User | false |
openclaw/openclaw | 4,299,549,718 | I_kwDOQb6kR88AAAABAEXsFg | 69,511 | https://github.com/openclaw/openclaw/issues/69511 | https://api.github.com/repos/openclaw/openclaw/issues/69511 | [Bug]: secrets audit PLAINTEXT_FOUND for synthetic codex-app-server apiKey in models.json | ## Product / Version
OpenClaw 2026.4.15 (041266a) — stable channel
## Summary
`openclaw secrets audit` flags the codex provider's `apiKey` field in every agent's `models.json` as `PLAINTEXT_FOUND`. The flagged value is `"codex-app-server"` — a synthetic auth marker hardcoded by the codex extension, not a real API ke... | closed | completed | true | 0 | [] | [] | 2026-04-21T00:45:29Z | 2026-04-24T04:26:13Z | 2026-04-21T11:38:28Z | NONE | null | 20260424T060021Z | 2026-04-24T06:00:21Z | joerod26 | 277,153,564 | U_kgDOEIUHHA | User | false |
openclaw/openclaw | 4,299,549,776 | I_kwDOQb6kR88AAAABAEXsUA | 69,513 | https://github.com/openclaw/openclaw/issues/69513 | https://api.github.com/repos/openclaw/openclaw/issues/69513 | [Bug]: cron create silently accepts invalid delivery accountId without validation | ## Product / Version
OpenClaw 2026.4.15 (041266a) — stable channel
## Summary
`openclaw cron create --announce --account <id>` does not validate the `accountId` against configured channel accounts. A cron job created with a nonexistent account name is silently accepted and stored with the bogus value. At delivery ti... | open | null | false | 1 | [] | [] | 2026-04-21T00:45:30Z | 2026-05-02T07:16:41Z | null | NONE | null | 20260502T090030Z | 2026-05-02T09:00:30Z | joerod26 | 277,153,564 | U_kgDOEIUHHA | User | false |
openclaw/openclaw | 4,299,549,750 | I_kwDOQb6kR88AAAABAEXsNg | 69,512 | https://github.com/openclaw/openclaw/issues/69512 | https://api.github.com/repos/openclaw/openclaw/issues/69512 | [Feature]: Forward exec-approvals.json allowlist to claude-cli backend sessions | ## Product / Version
OpenClaw 2026.4.15 (041266a) — stable channel
## Summary
When an agent uses `claude-cli` as its primary model provider, exec calls are handled internally by Claude Code. OpenClaw's `exec-approvals.json` allowlist and `tools.exec.security: "allowlist"` setting are never consulted. This creates tw... | open | null | false | 1 | [] | [] | 2026-04-21T00:45:30Z | 2026-05-02T20:57:34Z | null | NONE | null | 20260502T210030Z | 2026-05-02T21:00:30Z | joerod26 | 277,153,564 | U_kgDOEIUHHA | User | false |
openclaw/openclaw | 4,042,157,634 | I_kwDOQb6kR87w7m5C | 40,256 | https://github.com/openclaw/openclaw/issues/40256 | https://api.github.com/repos/openclaw/openclaw/issues/40256 | System prompt section ordering breaks LLM prefix caching for local models | ## Problem
`buildAgentSystemPrompt()` places dynamic, per-conversation content (Messaging/inline buttons, Group Chat Context, Reactions) **before** the large static Project Context block. This invalidates LLM prefix caching on every request where the channel context differs, causing 8-16x slower follow-up turns on loc... | closed | completed | true | 7 | [] | [] | 2026-03-08T20:56:04Z | 2026-05-01T04:41:07Z | 2026-04-28T23:50:46Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | rhclaw | 260,109,027 | U_kgDOD4Dy4w | User | false |
openclaw/openclaw | 4,299,685,727 | I_kwDOQb6kR88AAAABAEf_Xw | 69,523 | https://github.com/openclaw/openclaw/issues/69523 | https://api.github.com/repos/openclaw/openclaw/issues/69523 | [Feature]: Add option to auto-expand tool calls and tool outputs in Control UI | ### Summary
Add configuration option to auto-expand tool calls and tool outputs in Control UI by default
### Problem to solve
Currently, tool calls and tool outputs in Control UI are displayed as collapsible cards that require manual clicking to expand. This creates extra friction when:
Debugging tool executions
Re... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-04-21T01:27:31Z | 2026-04-29T04:34:37Z | 2026-04-26T18:57:23Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | rosyriver | 19,612,404 | MDQ6VXNlcjE5NjEyNDA0 | User | false |
openclaw/openclaw | 4,299,704,644 | I_kwDOQb6kR88AAAABAEhJRA | 69,525 | https://github.com/openclaw/openclaw/issues/69525 | https://api.github.com/repos/openclaw/openclaw/issues/69525 | Cross-provider message send leaks source session accountId to target channel | ## Bug Description
When using the `message` tool to send a cross-provider message (e.g., from a Telegram session to a WeChat/weixin channel), the source session's `accountId` (typically `"default"`) is passed to the target channel plugin's `sendText`/`sendMedia` handler.
## Root Cause
In `message-action-runner`, the... | open | null | false | 1 | [] | [] | 2026-04-21T01:32:20Z | 2026-05-02T13:08:07Z | null | NONE | null | 20260502T150030Z | 2026-05-02T15:00:30Z | iFwu | 16,523,864 | MDQ6VXNlcjE2NTIzODY0 | User | false |
openclaw/openclaw | 4,299,706,102 | I_kwDOQb6kR88AAAABAEhO9g | 69,526 | https://github.com/openclaw/openclaw/issues/69526 | https://api.github.com/repos/openclaw/openclaw/issues/69526 | [Bug]: CLI subcommand dispatch remains 60-180s on modern laptop hardware after #5871 fixes | # [Bug]: CLI subcommand dispatch remains 60–180s on modern laptop hardware after #5871 fixes
## Summary
After #5871 was closed with perf fixes merged, `openclaw <subcommand>` still takes **60–180 seconds** on a Core i7-10850H laptop (NVMe, 16 GB RAM). Reproduces on `2026.4.15`, a clean reinstall, and `2026.4.19-beta.... | closed | completed | true | 1 | [] | [] | 2026-04-21T01:32:44Z | 2026-04-29T04:42:42Z | 2026-04-26T06:58:53Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | Turtlesfr | 6,327,245 | MDQ6VXNlcjYzMjcyNDU= | User | false |
openclaw/openclaw | 4,299,726,655 | I_kwDOQb6kR88AAAABAEifPw | 69,527 | https://github.com/openclaw/openclaw/issues/69527 | https://api.github.com/repos/openclaw/openclaw/issues/69527 | [Bug]: openrouter/auto ignored by infer model run, falls back to openai/gpt-4.1-mini despite being configured as primary | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
### Version
OpenClaw 2026.4.15 (041266a)
### Environment
- Linux
- Gateway managed by `systemd`
- Service command:
`/usr/bin/node /home/alexifra/.local/lib/node_modules/openclaw/dist/index.js gateway --port 18789`
- Con... | open | null | false | 2 | [
"bug",
"regression"
] | [] | 2026-04-21T01:38:39Z | 2026-05-02T07:16:42Z | null | NONE | null | 20260502T090030Z | 2026-05-02T09:00:30Z | alexifra | 207,808,725 | U_kgDODGLo1Q | User | false |
openclaw/openclaw | 4,164,091,519 | I_kwDOQb6kR874Mv5_ | 57,020 | https://github.com/openclaw/openclaw/issues/57020 | https://api.github.com/repos/openclaw/openclaw/issues/57020 | [Bug]: Daily session reset creates new sessionId but preserves stale sessionFile — transcript never resets | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
When the daily session reset triggers (default 4AM), initSessionState() creates a new sessionId but the previous sessionFile path is carried over to the new session entry via resolveAndPersistSessionFile(), cau... | closed | completed | true | 2 | [
"bug"
] | [] | 2026-03-29T12:32:30Z | 2026-04-28T05:19:44Z | 2026-04-25T11:55:23Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | nailujac | 83,184,419 | MDQ6VXNlcjgzMTg0NDE5 | User | false |
openclaw/openclaw | 4,299,791,396 | I_kwDOQb6kR88AAAABAEmcJA | 69,531 | https://github.com/openclaw/openclaw/issues/69531 | https://api.github.com/repos/openclaw/openclaw/issues/69531 | [Feature]: Add staggerMs option to dreaming cron schedule | ## Summary
Allow configuring a random stagger window for the dreaming cron schedule so that multiple OpenClaw instances do not fire dreaming sweeps at the exact same time.
## Problem to solve
When running multiple OpenClaw instances (e.g., in a replicated or HA deployment), all instances sharing the same dreaming cr... | closed | not_planned | true | 1 | [] | [] | 2026-04-21T01:57:48Z | 2026-04-29T04:37:05Z | 2026-04-26T10:50:28Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | jackson-jia-914 | 252,321,914 | U_kgDODwogeg | User | false |
openclaw/openclaw | 4,299,797,320 | I_kwDOQb6kR88AAAABAEmzSA | 69,532 | https://github.com/openclaw/openclaw/issues/69532 | https://api.github.com/repos/openclaw/openclaw/issues/69532 | [Bug]: Browser tool cannot connect to self-hosted Browserless remote CDP in Docker even though /json/version?token=... succeeds | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
OpenClaw 2026.4.15 cannot establish a remote CDP browser connection to a self-hosted Browserless container in Docker, even though Browserless is r... | closed | completed | true | 1 | [
"bug"
] | [] | 2026-04-21T01:59:35Z | 2026-04-28T05:19:10Z | 2026-04-25T12:01:42Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Clawbert-bot | 261,691,954 | U_kgDOD5kaMg | User | false |
openclaw/openclaw | 4,299,800,519 | I_kwDOQb6kR88AAAABAEm_xw | 69,533 | https://github.com/openclaw/openclaw/issues/69533 | https://api.github.com/repos/openclaw/openclaw/issues/69533 | Feature Request: Make Dreaming System Prompt User-Customizable | **Problem**
The Dreaming system in OpenClaw has its prompt hardcoded in backend JavaScript (dreaming-*.js). Users cannot customize the language, section headers, or format of dream entries.
**Proposed Solution**
Add a `dreaming.prompt` config field in openclaw.json and expose it in Settings → Dreaming UI. Support vari... | open | null | false | 1 | [] | [] | 2026-04-21T02:00:30Z | 2026-05-02T17:31:37Z | null | NONE | null | 20260502T180029Z | 2026-05-02T18:00:29Z | mianayo | 105,649,803 | U_kgDOBkwWiw | User | false |
openclaw/openclaw | 4,299,814,168 | I_kwDOQb6kR88AAAABAEn1GA | 69,534 | https://github.com/openclaw/openclaw/issues/69534 | https://api.github.com/repos/openclaw/openclaw/issues/69534 | [Bug]: memory file naming format inconsistency between session-memory hook and dreaming system | %TSD-Header-###%$HtJ&=��~�h��E� � @ �=8˵ BP v��h?!P�v��o|ޚ2O�5�ڑ�L�!�q�j�����̉� o�ѥ])e�r�Ao=zM2�,�E�Q�jO�WRUm�*฿�!!��W!P�v��o|ޚ2�%�ڑ�M���j�����܉�
o�ѥM)e�v�Aa-zE�?K!P�� A||2 ... | closed | completed | true | 0 | [] | [] | 2026-04-21T02:04:26Z | 2026-04-23T04:22:11Z | 2026-04-21T02:06:26Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | cnnceart | 45,220,221 | MDQ6VXNlcjQ1MjIwMjIx | User | false |
openclaw/openclaw | 4,299,830,711 | I_kwDOQb6kR88AAAABAEo1tw | 69,536 | https://github.com/openclaw/openclaw/issues/69536 | https://api.github.com/repos/openclaw/openclaw/issues/69536 | [Bug]: memory file naming format inconsistency between session-memory hook and dreaming system | ## Bug Description
The session-memory hook and the dreaming/agent daily notes systems use different file naming conventions, causing memory files generated by one system to be invisible to the other.
## Two File Formats
### 1. Session-Memory Hook (/new or /reset)
- **Naming**: memory/YYYY-MM-DD-slug.md (e.g., 2026-... | open | null | false | 1 | [] | [] | 2026-04-21T02:08:51Z | 2026-05-02T12:54:03Z | null | NONE | null | 20260502T150030Z | 2026-05-02T15:00:30Z | cnnceart | 45,220,221 | MDQ6VXNlcjQ1MjIwMjIx | User | false |
openclaw/openclaw | 4,299,836,749 | I_kwDOQb6kR88AAAABAEpNTQ | 69,538 | https://github.com/openclaw/openclaw/issues/69538 | https://api.github.com/repos/openclaw/openclaw/issues/69538 | OpenRouter provider returns empty incomplete turn (stopReason=stop, payloads=0) with valid auth on 2026.4.15 | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
On OpenClaw 2026.4.15, OpenRouter-backed runs terminate with
[agent/embedded] incomplete turn detected: stopReason=stop payloads=0
and no assistant output. Reproduces in Control UI chat, openclaw agent, and ope... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-04-21T02:10:30Z | 2026-04-28T04:49:07Z | 2026-04-25T19:10:25Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | DAMI333N | 139,545,183 | U_kgDOCFFKXw | User | false |
openclaw/openclaw | 4,083,985,795 | I_kwDOQb6kR87zbK2D | 48,413 | https://github.com/openclaw/openclaw/issues/48413 | https://api.github.com/repos/openclaw/openclaw/issues/48413 | Bug: Cost calculation returns /bin/zsh for all API calls with moonshot/kimi-k2.5 | ## Bug Description
OpenClaw is reporting `/bin/zsh.0000` cost for all API calls when using `moonshot/kimi-k2.5` model, despite tokens being consumed and actual charges being incurred.
## Environment
- **OpenClaw Version:** 2026.3.13 (61d171a)
- **Model:** moonshot/kimi-k2.5
- **OS:** macOS Darwin 24.5.0 (arm64)
- **... | closed | completed | true | 3 | [] | [] | 2026-03-16T18:04:32Z | 2026-04-23T04:22:09Z | 2026-04-21T02:12:39Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | Wolverine971 | 26,643,835 | MDQ6VXNlcjI2NjQzODM1 | User | false |
openclaw/openclaw | 4,098,382,004 | I_kwDOQb6kR870SFi0 | 50,082 | https://github.com/openclaw/openclaw/issues/50082 | https://api.github.com/repos/openclaw/openclaw/issues/50082 | Bug: Cost Display Precision - Cached tokens show $0.00 instead of actual cost | ## Problem
When using Amazon Bedrock with 100% cache hit rate, tokens that cost < $0.01 display as `$0.00` instead of showing the actual cost (e.g., `$0.004` or `$0.005`).
This creates a misleading impression that cached tokens are completely free, when they actually have a reduced cost (90% discount on input tokens).... | closed | duplicate | true | 2 | [] | [] | 2026-03-19T00:01:04Z | 2026-04-23T04:22:09Z | 2026-04-21T02:12:50Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | tabotthebot | 264,255,103 | U_kgDOD8A2fw | User | false |
openclaw/openclaw | 4,098,382,108 | I_kwDOQb6kR870SFkc | 50,083 | https://github.com/openclaw/openclaw/issues/50083 | https://api.github.com/repos/openclaw/openclaw/issues/50083 | Bug: Cost Display Precision - Cached tokens show $0.00 instead of actual cost | ## Problem
When using Amazon Bedrock with 100% cache hit rate, tokens that cost < $0.01 display as `$0.00` instead of showing the actual cost (e.g., `$0.004` or `$0.005`).
This creates a misleading impression that cached tokens are completely free, when they actually have a reduced cost (90% discount on input tokens).... | closed | completed | true | 3 | [] | [] | 2026-03-19T00:01:07Z | 2026-04-23T04:22:08Z | 2026-04-21T02:12:57Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | tabotthebot | 264,255,103 | U_kgDOD8A2fw | User | false |
openclaw/openclaw | 4,299,860,015 | I_kwDOQb6kR88AAAABAEqoLw | 69,541 | https://github.com/openclaw/openclaw/issues/69541 | https://api.github.com/repos/openclaw/openclaw/issues/69541 | [RFC] Declarative registry for plugin-injected context XML tags (strip/sanitize/ingest contract) | # [RFC] Declarative registry for plugin-injected context XML tags (strip/sanitize/ingest contract)
## TLDR
Plugins that inject XML-tagged blocks into the prompt (`<hindsight_memories>`, `<relevant_memories>`, `<active_memory_plugin>`, etc.) are **invisible to downstream consumers** that need to strip them — compactio... | open | null | false | 1 | [] | [] | 2026-04-21T02:17:52Z | 2026-05-02T17:31:39Z | null | CONTRIBUTOR | null | 20260502T180029Z | 2026-05-02T18:00:29Z | jetd1 | 15,795,935 | MDQ6VXNlcjE1Nzk1OTM1 | User | false |
openclaw/openclaw | 4,299,910,376 | I_kwDOQb6kR88AAAABAEts6A | 69,543 | https://github.com/openclaw/openclaw/issues/69543 | https://api.github.com/repos/openclaw/openclaw/issues/69543 | Dreaming-generated sessions appear in normal Control UI session picker | ## Summary
Dreaming-generated sessions (for example `dreaming-narrative-light-*` and `dreaming-narrative-rem-*`) are being persisted into the normal session store and show up in the Control UI session switcher/dropdown as if they were ordinary user sessions.
This clutters the session picker with many internal/backgrou... | closed | completed | true | 1 | [] | [] | 2026-04-21T02:33:26Z | 2026-04-29T04:42:41Z | 2026-04-26T06:58:57Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | kiraje | 97,667,303 | U_kgDOBdJI5w | User | false |
openclaw/openclaw | 4,299,911,370 | I_kwDOQb6kR88AAAABAEtwyg | 69,544 | https://github.com/openclaw/openclaw/issues/69544 | https://api.github.com/repos/openclaw/openclaw/issues/69544 | [Feature]: Matrix — 允许通过配置开关控制 Agent emoji 是否触发 mention | ### Summary
Matrix 渠道中，Agent 的 `identity.emoji` 会被自动加入 mention 匹配模式，导致**任何包含该 emoji 的消息**都会触发 Agent 回复，等同于被 @。建议增加配置开关让用户自主决定是否启用 emoji 作为 mention 触发条件。
### Problem to solve
源码 `dist/mentions-D6uUpOuN.js` 中 `deriveMentionPatterns` 函数：
```js
function deriveMentionPatterns(identity) {
const patterns = [];
con... | open | null | false | 3 | [
"enhancement"
] | [] | 2026-04-21T02:33:50Z | 2026-05-02T14:36:38Z | null | NONE | null | 20260502T150030Z | 2026-05-02T15:00:30Z | lzn970721 | 62,925,906 | MDQ6VXNlcjYyOTI1OTA2 | User | false |
openclaw/openclaw | 4,299,924,495 | I_kwDOQb6kR88AAAABAEukDw | 69,546 | https://github.com/openclaw/openclaw/issues/69546 | https://api.github.com/repos/openclaw/openclaw/issues/69546 | [Bug]: QQBot channel does not apply updated bindings until gateway restart | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
QQBot routing does not reflect updated bindings at runtime; new bindings appear in CLI but are not evaluated until gateway restart.
### Steps to reproduce
1. Start OpenClaw 2026.4.15 with QQBot configured
2. ... | open | null | false | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-21T02:38:17Z | 2026-05-02T12:37:10Z | null | NONE | null | 20260502T150030Z | 2026-05-02T15:00:30Z | F32138 | 145,747,124 | U_kgDOCK_stA | User | false |
openclaw/openclaw | 4,299,926,174 | I_kwDOQb6kR88AAAABAEuqng | 69,547 | https://github.com/openclaw/openclaw/issues/69547 | https://api.github.com/repos/openclaw/openclaw/issues/69547 | [Bug]: cron TypeError: Cannot read properties of undefined (reading 'trim') | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
touch cron always show：TypeError: Cannot read properties of undefined (reading 'trim')
### Steps to reproduce
touch cron always show：TypeError: Cannot read properties of undefined (reading 'trim')
### Expected behavior
t... | closed | completed | true | 1 | [
"bug",
"close:already-fixed",
"dedupe:child",
"regression"
] | [
"vincentkoc"
] | 2026-04-21T02:38:57Z | 2026-04-28T05:29:06Z | 2026-04-25T09:37:10Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | kangkang121 | 50,684,475 | MDQ6VXNlcjUwNjg0NDc1 | User | false |
openclaw/openclaw | 4,299,937,934 | I_kwDOQb6kR88AAAABAEvYjg | 69,550 | https://github.com/openclaw/openclaw/issues/69550 | https://api.github.com/repos/openclaw/openclaw/issues/69550 | [Bug]: | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
图标挡住了openclaw回复的文本，严重影响阅读
<img width="304" height="211" alt="Image" src="https://github.com/user-attachments/assets/9bbb8017-4501-405b-8ce8-df1fed13f1ad" />
### Steps to reproduce
与openclaw对话即可
### Expected behavior
无
#... | closed | not_planned | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-21T02:43:21Z | 2026-04-29T04:34:36Z | 2026-04-26T18:57:28Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | YouYou-tol | 231,819,742 | U_kgDODdFJ3g | User | false |
openclaw/openclaw | 4,100,178,947 | I_kwDOQb6kR870Y8QD | 50,330 | https://github.com/openclaw/openclaw/issues/50330 | https://api.github.com/repos/openclaw/openclaw/issues/50330 | [Bug]: When you try to command :openclaw gateway start, there was nothing prompt and response | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
when i close my computer directly, i didn't close the windows PowerShell window, then when i restart my PC,
and try to turn on it, i command : openclaw gateway start, but nothing prompt and response.
then i use claude find the reason:
1. D... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-19T08:37:42Z | 2026-04-23T04:22:04Z | 2026-04-21T02:57:29Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | joryzhang | 69,902,372 | MDQ6VXNlcjY5OTAyMzcy | User | false |
openclaw/openclaw | 4,300,012,134 | I_kwDOQb6kR88AAAABAEz6Zg | 69,558 | https://github.com/openclaw/openclaw/issues/69558 | https://api.github.com/repos/openclaw/openclaw/issues/69558 | imageModel not resolved for agents with per-agent model override | When an agent has a `model` override (e.g. `ollama/glm-5.1:cloud`) but no per-agent `imageModel` (which is impossible because the schema has `additionalProperties: false` and doesn't include `imageModel` as a valid key on `agents.list.*`), the image tool fails to resolve a vision model.
The `agents.defaults.imageModel... | closed | completed | true | 1 | [] | [] | 2026-04-21T03:06:50Z | 2026-04-26T04:29:18Z | 2026-04-24T01:27:45Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | AlfredIrving | 260,044,957 | U_kgDOD3_4nQ | User | false |
openclaw/openclaw | 4,300,020,578 | I_kwDOQb6kR88AAAABAE0bYg | 69,561 | https://github.com/openclaw/openclaw/issues/69561 | https://api.github.com/repos/openclaw/openclaw/issues/69561 | [Docs]: warn that granting macOS Accessibility to node grants GUI automation to all npm packages | ## Summary
OpenClaw on macOS can rely on Peekaboo or similar GUI automation paths that lead users to grant Accessibility permissions to the `node` executable. That is a meaningful security risk and should be documented prominently.
Granting Accessibility to `node` does not scope the permission to OpenClaw. It effecti... | closed | completed | true | 2 | [] | [] | 2026-04-21T03:09:07Z | 2026-04-28T04:49:05Z | 2026-04-25T19:10:38Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Shadow-of-Silence | 277,936,267 | U_kgDOEJD4iw | User | false |
openclaw/openclaw | 4,300,046,089 | I_kwDOQb6kR88AAAABAE1_CQ | 69,569 | https://github.com/openclaw/openclaw/issues/69569 | https://api.github.com/repos/openclaw/openclaw/issues/69569 | Session closure metadata leaking into user-visible messages | **Problem:**
When a long-running session ends and a new one starts, internal session closure messages are being appended to user messages and displayed in the UI.
**Example message user sees:**
```
SESSION FULLY ARCHIVED — Extended session April 14-20, 2026 (200K+ tokens) permanently closed. All systems stable. Ready ... | closed | not_planned | true | 1 | [] | [] | 2026-04-21T03:18:12Z | 2026-04-28T04:49:04Z | 2026-04-25T19:10:44Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | AIdenB899 | 269,310,343 | U_kgDOEA1Zhw | User | false |
openclaw/openclaw | 4,300,044,591 | I_kwDOQb6kR88AAAABAE15Lw | 69,565 | https://github.com/openclaw/openclaw/issues/69565 | https://api.github.com/repos/openclaw/openclaw/issues/69565 | [Bug] Async command cascade creates unstoppable approval loop | ## Description
**Severity**: Critical
**Reproducibility**: Consistent when multiple exec commands are approved in quick succession
**Session**: 2026-04-21, 08:39-09:34 GMT+8, webchat channel
---
### Problem Statement
When a user approves multiple `exec` commands in quick succession, each approval generates a se... | closed | not_planned | true | 1 | [] | [] | 2026-04-21T03:17:39Z | 2026-04-23T04:22:03Z | 2026-04-21T03:18:33Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | xiaobu1112 | 267,266,960 | U_kgDOD-4rkA | User | false |
openclaw/openclaw | 4,300,044,657 | I_kwDOQb6kR88AAAABAE15cQ | 69,566 | https://github.com/openclaw/openclaw/issues/69566 | https://api.github.com/repos/openclaw/openclaw/issues/69566 | [Bug] Async command cascade creates unstoppable approval loop | ## Description
**Severity**: Critical
**Reproducibility**: Consistent when multiple exec commands are approved in quick succession
**Session**: 2026-04-21, 08:39-09:34 GMT+8, webchat channel
---
### Problem Statement
When a user approves multiple `exec` commands in quick succession, each approval generates a se... | closed | not_planned | true | 1 | [] | [] | 2026-04-21T03:17:41Z | 2026-04-23T04:22:02Z | 2026-04-21T03:20:06Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | xiaobu1112 | 267,266,960 | U_kgDOD-4rkA | User | false |
openclaw/openclaw | 4,300,045,121 | I_kwDOQb6kR88AAAABAE17QQ | 69,568 | https://github.com/openclaw/openclaw/issues/69568 | https://api.github.com/repos/openclaw/openclaw/issues/69568 | [Bug] User 'deny' on exec approval is misinterpreted as 'try alternative approach' instead of 'stop', creating persistent cross-task behavioral drift | ## Description
**Severity**: High
**Reproducibility**: Consistent
**Session**: 2026-04-21, 08:39-09:34 GMT+8, webchat channel
---
### Problem Statement
When a user sends `/approve <id> deny` to stop a command execution, the agent interprets the denial as "this approach failed, try an alternative method" instead... | closed | not_planned | true | 1 | [] | [] | 2026-04-21T03:17:51Z | 2026-04-23T04:22:02Z | 2026-04-21T03:20:08Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | xiaobu1112 | 267,266,960 | U_kgDOD-4rkA | User | false |
openclaw/openclaw | 4,300,039,602 | I_kwDOQb6kR88AAAABAE1lsg | 69,563 | https://github.com/openclaw/openclaw/issues/69563 | https://api.github.com/repos/openclaw/openclaw/issues/69563 | [Bug] Async command cascade creates unstoppable approval loop | {
"title": "[Bug] Async command cascade creates unstoppable approval loop — each approval generates delayed completion message, user commands blocked until all fire",
"body": "## Description\n\n**Severity**: Critical \n**Reproducibility**: Consistent when multiple exec commands are approved in quick succession \n... | closed | not_planned | true | 1 | [] | [] | 2026-04-21T03:15:51Z | 2026-04-23T04:22:01Z | 2026-04-21T03:22:53Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | xiaobu1112 | 267,266,960 | U_kgDOD-4rkA | User | false |
openclaw/openclaw | 4,300,059,358 | I_kwDOQb6kR88AAAABAE2y3g | 69,570 | https://github.com/openclaw/openclaw/issues/69570 | https://api.github.com/repos/openclaw/openclaw/issues/69570 | [Bug] Async command cascade creates unstoppable approval loop — each approval generates delayed completion message, user commands blocked until all fire | ## Description
**Severity**: Critical
**Reproducibility**: Consistent when multiple exec commands are approved in quick succession
**Session**: 2026-04-21, 08:39-09:34 GMT+8, webchat channel
---
### Problem Statement
When a user approves multiple `exec` commands in quick succession, each approval generates a se... | open | null | false | 2 | [] | [] | 2026-04-21T03:23:14Z | 2026-05-02T16:55:41Z | null | NONE | null | 20260502T180029Z | 2026-05-02T18:00:29Z | xiaobu1112 | 267,266,960 | U_kgDOD-4rkA | User | false |
openclaw/openclaw | 4,300,042,842 | I_kwDOQb6kR88AAAABAE1yWg | 69,564 | https://github.com/openclaw/openclaw/issues/69564 | https://api.github.com/repos/openclaw/openclaw/issues/69564 | [Bug] Async command cascade creates unstoppable approval loop — each approval generates delayed completion message, user commands blocked until all fire | ## Description
**Severity**: Critical
**Reproducibility**: Consistent when multiple exec commands are approved in quick succession
**Session**: 2026-04-21, 08:39-09:34 GMT+8, webchat channel
---
### Problem Statement
When a user approves multiple `exec` commands in quick succession, each approval generates a se... | closed | not_planned | true | 1 | [] | [] | 2026-04-21T03:17:00Z | 2026-04-23T04:22:00Z | 2026-04-21T03:25:20Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | xiaobu1112 | 267,266,960 | U_kgDOD-4rkA | User | false |
openclaw/openclaw | 4,300,065,673 | I_kwDOQb6kR88AAAABAE3LiQ | 69,572 | https://github.com/openclaw/openclaw/issues/69572 | https://api.github.com/repos/openclaw/openclaw/issues/69572 | [Bug]: Feishu typing indicator should use Typing API instead of Message Reaction API | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
The Feishu channel's typing indicator is currently implemented using im.messageReaction.create API (adding an emoji reaction with emoji_type: "Typing"), but this doesn't actually display the native "typing..." ... | open | null | false | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-21T03:25:14Z | 2026-05-02T16:01:53Z | null | NONE | null | 20260502T180029Z | 2026-05-02T18:00:29Z | biaochegg | 13,598,783 | MDQ6VXNlcjEzNTk4Nzgz | User | false |
openclaw/openclaw | 4,300,066,987 | I_kwDOQb6kR88AAAABAE3Qqw | 69,573 | https://github.com/openclaw/openclaw/issues/69573 | https://api.github.com/repos/openclaw/openclaw/issues/69573 | [Bug] User 'deny' on exec approval is misinterpreted as 'try alternative approach' instead of 'stop', creating persistent cross-task behavioral drift | ## Description
**Severity**: High
**Reproducibility**: Consistent
**Session**: 2026-04-21, 08:39-09:34 GMT+8, webchat channel
---
### Problem Statement
When a user sends `/approve <id> deny` to stop a command execution, the agent interprets the denial as "this approach failed, try an alternative method" instead... | closed | completed | true | 1 | [] | [] | 2026-04-21T03:25:43Z | 2026-04-28T04:49:04Z | 2026-04-25T19:10:51Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | xiaobu1112 | 267,266,960 | U_kgDOD-4rkA | User | false |
openclaw/openclaw | 4,300,088,437 | I_kwDOQb6kR88AAAABAE4kdQ | 69,576 | https://github.com/openclaw/openclaw/issues/69576 | https://api.github.com/repos/openclaw/openclaw/issues/69576 | [Bug]: Delivery-mirror leaks messages across agents when multiple Telegram accounts bind to same peer | ## Summary
When multiple agents are bound to the same Telegram peer via different Telegram accounts (different bot tokens), the `delivery-mirror` mechanism incorrectly syncs outbound messages to **all** sessions bound to that peer, instead of only the session belonging to the agent that sent the message.
## Environme... | closed | completed | true | 1 | [] | [] | 2026-04-21T03:32:03Z | 2026-04-28T04:49:03Z | 2026-04-25T19:10:57Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | huyg0415 | 134,746,285 | U_kgDOCAgQrQ | User | false |
openclaw/openclaw | 4,300,119,583 | I_kwDOQb6kR88AAAABAE6eHw | 69,579 | https://github.com/openclaw/openclaw/issues/69579 | https://api.github.com/repos/openclaw/openclaw/issues/69579 | [Bug]: Telegram handler drops thinking block signatures on multi-turn conversations, causing Anthropic API rejection | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Multi-turn Telegram conversations with claude-opus-4-7 (extended thinking enabled) fail on turn 2+ with "messages.1.content.0: Invalid signature in thinking block" because OpenClaw's Telegram handler does not p... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-21T03:42:56Z | 2026-04-28T04:49:02Z | 2026-04-25T19:11:11Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | jamesvblian | 258,092,914 | U_kgDOD2Ivcg | User | false |
openclaw/openclaw | 4,300,139,118 | I_kwDOQb6kR88AAAABAE7qbg | 69,582 | https://github.com/openclaw/openclaw/issues/69582 | https://api.github.com/repos/openclaw/openclaw/issues/69582 | Parameter injection "{}" causes an infinite loop in tool invocation | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
When calling the local model glm-5 tool, the parameters are inevitably 'contaminated' by `{}`
## Affected Tools
| Tool | Parameter | Expected Type | Corrupted Value |
|------|-----------|---------------|------... | open | null | false | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-04-21T03:48:36Z | 2026-05-02T17:31:40Z | null | NONE | null | 20260502T180029Z | 2026-05-02T18:00:29Z | ZKANGGIT | 36,831,988 | MDQ6VXNlcjM2ODMxOTg4 | User | false |
openclaw/openclaw | 4,300,233,967 | I_kwDOQb6kR88AAAABAFBc7w | 69,589 | https://github.com/openclaw/openclaw/issues/69589 | https://api.github.com/repos/openclaw/openclaw/issues/69589 | [Bug]: Bun runtime crashes on bootstrap — isDirectModuleNotFoundError does not recognize Bun error shape | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Running `bun openclaw.mjs` crashes immediately with `Cannot find module './dist/warning-filter.js'` on v2026.3.22 and later. The `isDirectModuleNotFoundError` function in `openclaw.mjs` relies on Node-specific error properti... | closed | not_planned | true | 1 | [] | [] | 2026-04-21T04:15:43Z | 2026-04-29T04:42:40Z | 2026-04-26T07:00:44Z | CONTRIBUTOR | null | 20260429T060025Z | 2026-04-29T06:00:25Z | zhumengzhu | 4,508,623 | MDQ6VXNlcjQ1MDg2MjM= | User | false |
openclaw/openclaw | 3,946,560,008 | I_kwDOQb6kR87rO7oI | 17,890 | https://github.com/openclaw/openclaw/issues/17890 | https://api.github.com/repos/openclaw/openclaw/issues/17890 | [Bug]: macOS app: Skill binary detection doesn't respect `/etc/paths` or Homebrew PATH | ### Summary
The macOS app's skill page shows "Missing binaries: peekaboo" even though `peekaboo` is installed via Homebrew at `/opt/homebrew/bin/peekaboo`.
The app's process PATH is hardcoded to `/usr/bin:/bin:/usr/sbin:/sbin`, which excludes both `/opt/homebrew/bin` (Homebrew on Apple Silicon) and `/usr/local/bin` (... | closed | completed | true | 15 | [
"bug"
] | [] | 2026-02-16T08:37:23Z | 2026-04-27T04:29:30Z | 2026-04-24T19:03:43Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | cliffchen | 7,713,792 | MDQ6VXNlcjc3MTM3OTI= | User | false |
openclaw/openclaw | 3,973,464,747 | I_kwDOQb6kR87s1kKr | 23,208 | https://github.com/openclaw/openclaw/issues/23208 | https://api.github.com/repos/openclaw/openclaw/issues/23208 | TTS tool ignores messages.tts.elevenlabs.voiceId config | ## Bug
The built-in `tts` tool does not use the `voiceId` configured in `messages.tts.elevenlabs.voiceId` in `openclaw.json`. It appears to fall back to a default voice regardless of the config.
## Steps to Reproduce
1. Configure ElevenLabs TTS in `openclaw.json`:
```json
{
"messages": {
"tts": {
"provid... | closed | not_planned | true | 12 | [
"stale"
] | [] | 2026-02-22T03:38:45Z | 2026-04-27T05:04:21Z | 2026-04-24T19:06:44Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | mcdeeai | 2,171,421 | MDQ6VXNlcjIxNzE0MjE= | User | false |
openclaw/openclaw | 3,997,239,388 | I_kwDOQb6kR87uQQhc | 27,851 | https://github.com/openclaw/openclaw/issues/27851 | https://api.github.com/repos/openclaw/openclaw/issues/27851 | [Bug]:Discord EventQueue timeout (30s) too aggressive for AI responses | ### Summary
# OpenClaw Discord EventQueue Timeout Bug Report
## Issue Summary
Discord listener consistently times out after 30 seconds, causing the bot to stop responding to messages.
## Problem Description
The Discord EventQueue has a hardcoded 30-second timeout that is too aggressive for AI responses. When a respo... | closed | completed | true | 9 | [
"bug",
"stale"
] | [] | 2026-02-26T19:00:08Z | 2026-04-27T05:04:00Z | 2026-04-24T19:08:34Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | thebotapex-coder | 264,241,747 | U_kgDOD8ACUw | User | false |
openclaw/openclaw | 4,002,390,603 | I_kwDOQb6kR87uj6JL | 29,053 | https://github.com/openclaw/openclaw/issues/29053 | https://api.github.com/repos/openclaw/openclaw/issues/29053 | [Feature Request] MCP Client: Native support for connecting to external MCP servers | ## Summary
Add native MCP (Model Context Protocol) client support to OpenClaw, so agents can connect to and use any standard MCP server without workarounds.
## Problem
OpenClaw has its own built-in tool system which works well, but it is OpenClaw-specific. MCP is rapidly becoming the industry standard — supported by... | closed | completed | true | 18 | [
"stale"
] | [] | 2026-02-27T18:49:44Z | 2026-04-27T05:03:53Z | 2026-04-24T19:09:11Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | 4shil | 166,588,383 | U_kgDOCe3v3w | User | false |
openclaw/openclaw | 4,003,293,633 | I_kwDOQb6kR87unWnB | 29,214 | https://github.com/openclaw/openclaw/issues/29214 | https://api.github.com/repos/openclaw/openclaw/issues/29214 | [Bug]: Signal doesn't work in group chat with @ mention [Workaround provided] | ### Summary
Currently, openclaw will reply EVERY single message within a chat group, which make it not usable in a group chat, and @mention doesn't work because openclaw.json doesn't support the group tag for Signal
"groups": {
"*": {
"requireMention": true
}
},
### Steps to reproduce... | closed | completed | true | 15 | [
"bug"
] | [] | 2026-02-27T23:32:24Z | 2026-04-27T05:03:50Z | 2026-04-24T19:09:23Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | KritBlade | 256,668,062 | U_kgDOD0xxng | User | false |
openclaw/openclaw | 4,003,767,129 | I_kwDOQb6kR87upKNZ | 29,387 | https://github.com/openclaw/openclaw/issues/29387 | https://api.github.com/repos/openclaw/openclaw/issues/29387 | [Bug]: Bootstrap files in agentDir are silently ignored — only workspace directory files are injected into system prompt | ### Summary
When per-agent `agentDir` is configured (e.g. `~/.openclaw/agents/main/agent/`), placing bootstrap `.md` files (SOUL.md, AGENTS.md, TOOLS.md, USER.md) inside that directory has **no effect**. Only files under the shared `workspace` directory (`~/.openclaw/workspace/`) are loaded into the system prompt. The... | open | null | false | 13 | [
"bug"
] | [] | 2026-02-28T03:31:08Z | 2026-05-05T02:56:26Z | null | NONE | null | 20260505T030055Z | 2026-05-05T03:00:55Z | tuna-chin | 34,429,777 | MDQ6VXNlcjM0NDI5Nzc3 | User | false |
openclaw/openclaw | 4,010,503,151 | I_kwDOQb6kR87vC2vv | 31,583 | https://github.com/openclaw/openclaw/issues/31583 | https://api.github.com/repos/openclaw/openclaw/issues/31583 | [Bug]: `exec` tool does not inherit `skills.entries.*.env` environment variables | ### Bug type
Regression (worked before, now fails)
### Summary
## Summary
Environment variables configured under `skills.entries.<skill>.env` in `openclaw.json` are not passed through to subprocesses spawned by the `exec` tool. This makes it impossible to inject secrets (e.g. `GOG_KEYRING_PASSWORD`) into CLI tools ... | open | null | false | 11 | [
"bug",
"regression"
] | [] | 2026-03-02T10:32:35Z | 2026-05-05T02:56:39Z | null | NONE | null | 20260505T030055Z | 2026-05-05T03:00:55Z | cwebb77 | 101,180,764 | U_kgDOBgflXA | User | false |
openclaw/openclaw | 4,015,754,752 | I_kwDOQb6kR87vW44A | 32,868 | https://github.com/openclaw/openclaw/issues/32868 | https://api.github.com/repos/openclaw/openclaw/issues/32868 | Block streaming: block replies not delivered before tool execution (same-channel) | ## Summary
When block streaming is enabled with `blockStreamingBreak: "text_end"`, text chunks between tool calls are not delivered to the channel in real-time. Instead, all block replies arrive together after the model turn completes, defeating the purpose of progressive output during multi-tool-call turns.
## Expec... | open | null | false | 6 | [] | [] | 2026-03-03T09:57:51Z | 2026-05-05T02:56:50Z | null | CONTRIBUTOR | null | 20260505T030055Z | 2026-05-05T03:00:55Z | nxmxbbd | 32,288 | MDQ6VXNlcjMyMjg4 | User | false |
openclaw/openclaw | 4,017,340,728 | I_kwDOQb6kR87vc8E4 | 33,185 | https://github.com/openclaw/openclaw/issues/33185 | https://api.github.com/repos/openclaw/openclaw/issues/33185 | [Bug]: Critical Bug: imageModel fails with "Unknown model" despite valid config | ### Bug type
Regression (worked before, now fails)
### Summary
The 'imageModel' feature is completely broken, returning "Unknown model" for any valid, externally-verified visual model configuration.
### Steps to reproduce
1. **Configure `openclaw.json`** with valid API Keys for multiple providers that offer vision... | closed | completed | true | 12 | [
"bug",
"dedupe:parent",
"regression"
] | [] | 2026-03-03T15:36:46Z | 2026-04-28T04:45:48Z | 2026-04-25T23:08:29Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Kobe9312 | 201,453,280 | U_kgDODAHu4A | User | false |
openclaw/openclaw | 4,300,255,670 | I_kwDOQb6kR88AAAABAFCxtg | 69,591 | https://github.com/openclaw/openclaw/issues/69591 | https://api.github.com/repos/openclaw/openclaw/issues/69591 | Feature: URL routing rules via before_tool_call hook — redirect x.com/twitter.com to skill instead of web_fetch | ## Summary
When a user drops an X.com (or any domain) URL, the agent reflexively calls `web_fetch`. For X.com this always fails — X blocks unauthenticated scrapers. The right path is to use the `xread` skill (Grok API), but no enforcement mechanism exists today. Text rules in `AGENTS.md` aren't reliable enough because... | closed | not_planned | true | 1 | [] | [] | 2026-04-21T04:22:32Z | 2026-04-28T04:49:01Z | 2026-04-25T19:11:18Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Bartok9 | 259,807,879 | U_kgDOD3xahw | User | false |
openclaw/openclaw | 4,028,853,722 | I_kwDOQb6kR87wI23a | 36,323 | https://github.com/openclaw/openclaw/issues/36323 | https://api.github.com/repos/openclaw/openclaw/issues/36323 | Telegram: raw markdown tables render as wall of text — needs platform-level formatting gate | ## Problem
Markdown tables sent to Telegram render as a wall of text with raw pipes and dashes instead of formatted tables. This is a known, persistent issue affecting multiple users (see [Discussion #26825](https://github.com/openclaw/openclaw/discussions/26825)).
**What the user sees on Telegram:**
```
🎯 Status |... | closed | completed | true | 8 | [] | [] | 2026-03-05T15:09:29Z | 2026-04-27T05:03:04Z | 2026-04-24T19:24:10Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | sene1337 | 261,265,124 | U_kgDOD5KW5A | User | false |
openclaw/openclaw | 4,034,904,524 | I_kwDOQb6kR87wf8HM | 38,067 | https://github.com/openclaw/openclaw/issues/38067 | https://api.github.com/repos/openclaw/openclaw/issues/38067 | [Bug]: Cron isolated sessions lose sandbox.docker.dangerouslyAllow* flags due to shallow Object.assign on agents.defaults | ## Bug Description
`runCronIsolatedAgentTurn()` in `src/cron/isolated-agent/run.ts` builds a modified config by shallow-merging the per-agent config on top of `agents.defaults`:
```typescript
const { model: overrideModel, ...agentOverrideRest } = agentConfigOverride ?? {};
const agentCfg: AgentDefaultsConfig = Object... | closed | completed | true | 6 | [
"stale"
] | [] | 2026-03-06T15:52:21Z | 2026-04-27T05:02:05Z | 2026-04-24T19:30:05Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | accek-bot | 263,030,748 | U_kgDOD62H3A | User | false |
openclaw/openclaw | 4,034,924,355 | I_kwDOQb6kR87wgA9D | 38,076 | https://github.com/openclaw/openclaw/issues/38076 | https://api.github.com/repos/openclaw/openclaw/issues/38076 | skill-creator: make init_skill --resources parsing case-insensitive | ## Summary
`skills/skill-creator/scripts/init_skill.py` currently requires exact lowercase values for `--resources` (for example `scripts,references,assets`).
In real usage, mixed-case values are common (`Scripts,References,ASSETS`) and are semantically equivalent, but they are rejected today.
## Proposed improvement... | open | null | false | 3 | [] | [] | 2026-03-06T15:56:39Z | 2026-05-04T06:03:38Z | null | CONTRIBUTOR | null | 20260504T090105Z | 2026-05-04T09:01:05Z | shuofengzhang | 24,763,026 | MDQ6VXNlcjI0NzYzMDI2 | User | false |
openclaw/openclaw | 4,034,931,027 | I_kwDOQb6kR87wgClT | 38,077 | https://github.com/openclaw/openclaw/issues/38077 | https://api.github.com/repos/openclaw/openclaw/issues/38077 | gmail channel: allowFrom ["*"] semantics unclear — causes unintended auto-replies to all senders | ## Problem
The `allowFrom` field in the Gmail channel config has fundamentally different semantics than in other channels (Telegram, Discord), but this is not documented or warned about.
In Telegram/Discord: `allowFrom: ["*"]` means "accept messages from all users" — purely a receive filter.
In Gmail: `allowFrom: ["... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-06T15:58:13Z | 2026-04-27T05:02:04Z | 2026-04-24T19:30:10Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | sabo961 | 191,984 | MDQ6VXNlcjE5MTk4NA== | User | false |
openclaw/openclaw | 4,034,932,056 | I_kwDOQb6kR87wgC1Y | 38,078 | https://github.com/openclaw/openclaw/issues/38078 | https://api.github.com/repos/openclaw/openclaw/issues/38078 | Feature: Seamless session continuity across gateway restarts (client reconnect + thread resume) | ## Problem
When the gateway restarts, **the session context survives** (conversation history is persisted, the agent can resume), but the **active client connection drops** and the user must manually re-prompt to pick the thread back up.
This creates a disruptive experience for operators using webchat, TUI, or any pe... | closed | completed | true | 2 | [
"stale"
] | [] | 2026-03-06T15:58:25Z | 2026-04-27T05:02:03Z | 2026-04-24T19:30:15Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Adam-Researchh | 261,827,313 | U_kgDOD5sq8Q | User | false |
openclaw/openclaw | 4,034,952,927 | I_kwDOQb6kR87wgH7f | 38,083 | https://github.com/openclaw/openclaw/issues/38083 | https://api.github.com/repos/openclaw/openclaw/issues/38083 | [Bug]: Slack channel auth failures cause retry storm without backoff, degrading other channels | ## Summary
When a Slack token expires or becomes invalid, the gateway enters a tight retry loop (~1.5s intervals) that continues indefinitely without exponential backoff. While the gateway does not crash, this floods logs and can degrade access to other channels (webchat, TUI).
**Related issues:**
- #21082 — Slack So... | closed | completed | true | 3 | [
"stale"
] | [] | 2026-03-06T16:02:37Z | 2026-04-27T05:02:03Z | 2026-04-24T19:30:25Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | abndnt | 201,772,587 | U_kgDODAbOKw | User | false |
openclaw/openclaw | 4,034,970,750 | I_kwDOQb6kR87wgMR- | 38,090 | https://github.com/openclaw/openclaw/issues/38090 | https://api.github.com/repos/openclaw/openclaw/issues/38090 | [Bug] OpenClaw UI WebSocket reconnect 导致会话 terminated | ## 问题描述
OpenClaw UI 的 WebSocket 连接频繁触发 `reconnect`，导致正在进行的 AI 会话被 `terminated`。
## 环境
- OpenClaw 版本: 2026.3.2 (85377a2)
- 运行环境: WSL2 on Windows 11
- Node.js: v22.22.1
- 访问方式: OpenClaw UI (webchat) via localhost
## 复现步骤
1. 在 OpenClaw UI 中与 AI 进行对话
2. 等待一段时间（无需任何操作）
3. WebSocket 自动触发 reconnect
4. 如果此时正好有 AI 会话在处理，会话... | closed | not_planned | true | 2 | [] | [] | 2026-03-06T16:06:07Z | 2026-04-29T04:38:05Z | 2026-04-26T10:02:42Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | tociyou1995 | 150,312,743 | U_kgDOCPWXJw | User | false |
openclaw/openclaw | 4,034,972,482 | I_kwDOQb6kR87wgMtC | 38,091 | https://github.com/openclaw/openclaw/issues/38091 | https://api.github.com/repos/openclaw/openclaw/issues/38091 | [Bug] OpenClaw UI WebSocket reconnect 导致会话 terminated | ## 问题描述
OpenClaw UI 的 WebSocket 连接频繁触发 `reconnect`，导致正在进行的 AI 会话被 `terminated`。
## 环境
- OpenClaw 版本: 2026.3.2 (85377a2)
- 运行环境: WSL2 on Windows 11
- Node.js: v22.22.1
- 访问方式: OpenClaw UI (webchat) via localhost
## 复现步骤
1. 在 OpenClaw UI 中与 AI 进行对话
2. 等待一段时间（无需任何操作）
3. WebSocket 自动触发 reconnect
4. 如果此时正好有 AI 会话在处理，会话... | open | null | false | 4 | [] | [] | 2026-03-06T16:06:29Z | 2026-05-04T06:03:39Z | null | NONE | null | 20260504T090105Z | 2026-05-04T09:01:05Z | tociyou1995 | 150,312,743 | U_kgDOCPWXJw | User | false |
openclaw/openclaw | 4,035,001,882 | I_kwDOQb6kR87wgT4a | 38,120 | https://github.com/openclaw/openclaw/issues/38120 | https://api.github.com/repos/openclaw/openclaw/issues/38120 | Compaction fails with 'Unknown model: openai-codex/gpt-5.4-pro' | ## Description
Compaction fails with `Unknown model: openai-codex/gpt-5.4-pro` even though the model is working correctly for inference via the openai-codex OAuth provider.
## Environment
- OpenClaw version: 2026.3.2 (85377a2)
- Model: openai-codex/gpt-5.4-pro (GPT-5.4 Pro, released 2026-03-05)
- Auth: OAuth (openai-c... | closed | completed | true | 3 | [
"stale"
] | [] | 2026-03-06T16:13:23Z | 2026-04-27T05:09:43Z | 2026-04-24T05:22:24Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | jsquared71 | 4,314,939 | MDQ6VXNlcjQzMTQ5Mzk= | User | false |
openclaw/openclaw | 4,035,015,829 | I_kwDOQb6kR87wgXSV | 38,124 | https://github.com/openclaw/openclaw/issues/38124 | https://api.github.com/repos/openclaw/openclaw/issues/38124 | [Bug]: Node host cannot pair: "pairing required" while "openclaw nodes pending" is always empty (Oracle Cloud Ubuntu) | ### Bug type
Regression (worked before, now fails)
### Summary
## Summary
On an Oracle Cloud Ubuntu VPS, `openclaw node run` consistently fails with `pairing required`, but `openclaw nodes pending` always returns no pending requests.
This prevents approving the node and blocks node-host execution flow.
## Environme... | closed | completed | true | 2 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-06T16:16:42Z | 2026-04-27T05:02:02Z | 2026-04-24T19:30:30Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | nzf210 | 75,715,283 | MDQ6VXNlcjc1NzE1Mjgz | User | false |
openclaw/openclaw | 4,035,038,122 | I_kwDOQb6kR87wgcuq | 38,133 | https://github.com/openclaw/openclaw/issues/38133 | https://api.github.com/repos/openclaw/openclaw/issues/38133 | CC Chat: /usage footer and responseUsage line not appearing in agent replies | ## Bug
When `/usage tokens` or `/usage full` is enabled on a CC Chat session, the usage footer line (e.g. `📊 Usage: 1.2k in / 890 out`) does **not** appear on agent replies. It only works for pre-agent command replies (like the `/usage` confirmation itself).
## Root Cause
Two different text paths exist for CC Chat:... | closed | completed | true | 1 | [
"stale"
] | [] | 2026-03-06T16:22:06Z | 2026-04-28T04:49:39Z | 2026-04-25T19:04:51Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | fuller-stack-dev | 263,060,202 | U_kgDOD6366g | User | false |
openclaw/openclaw | 4,035,065,292 | I_kwDOQb6kR87wgjXM | 38,138 | https://github.com/openclaw/openclaw/issues/38138 | https://api.github.com/repos/openclaw/openclaw/issues/38138 | Feature request: mention-based agent routing override | ## Problem
In a multi-agent Discord setup, threads inherit their parent channel's agent binding (`parentPeer` match, routing tier 2). When a user @mentions a different agent in that thread, the mention is used only for **gating** (should the bot respond?), not for **routing** (which agent handles it).
**Example:** `#... | open | null | false | 2 | [] | [] | 2026-03-06T16:28:41Z | 2026-05-04T09:03:05Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | dbdbdipdip | 258,109,600 | U_kgDOD2JwoA | User | false |
openclaw/openclaw | 4,035,077,241 | I_kwDOQb6kR87wgmR5 | 38,141 | https://github.com/openclaw/openclaw/issues/38141 | https://api.github.com/repos/openclaw/openclaw/issues/38141 | Discord thread-bound subagent session binding fails (session mode unavailable) | null | closed | completed | true | 3 | [
"stale"
] | [] | 2026-03-06T16:31:31Z | 2026-04-27T04:54:59Z | 2026-04-24T22:05:37Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | BfreeIsMe | 200,308,618 | U_kgDOC_B3ig | User | false |
openclaw/openclaw | 4,035,087,929 | I_kwDOQb6kR87wgo45 | 38,145 | https://github.com/openclaw/openclaw/issues/38145 | https://api.github.com/repos/openclaw/openclaw/issues/38145 | maxConcurrent setting has no effect on Telegram message processing (code-splitting issue) | ## Bug Description
`agents.defaults.maxConcurrent` has no effect on inbound Telegram message processing. All Telegram messages are processed with an effective concurrency of 1 regardless of the configured value.
## Root Cause
The Rolldown bundler duplicates the queue code (`lanes` Map, `getLaneState`, `enqueueComman... | closed | completed | true | 4 | [
"stale"
] | [] | 2026-03-06T16:33:41Z | 2026-04-27T05:02:01Z | 2026-04-24T19:30:50Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | ibob92 | 24,282,053 | MDQ6VXNlcjI0MjgyMDUz | User | false |
openclaw/openclaw | 4,035,162,146 | I_kwDOQb6kR87wg7Ai | 38,166 | https://github.com/openclaw/openclaw/issues/38166 | https://api.github.com/repos/openclaw/openclaw/issues/38166 | Bootstrap agent attempts to write local files via messaging tool | During initial identity setup, the agent attempted to update local configuration files (IDENTITY.md, USER.md) by calling the message tool with absolute file paths. This caused Telegram API errors as the provider treated paths as chat IDs. Additionally, a '👿' emoji reaction failed with REACTION_INVALID (400 Bad Request... | closed | completed | true | 4 | [
"stale"
] | [] | 2026-03-06T16:49:09Z | 2026-04-27T05:02:00Z | 2026-04-24T19:30:59Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | oneles | 62,010,096 | MDQ6VXNlcjYyMDEwMDk2 | User | false |
openclaw/openclaw | 4,035,168,341 | I_kwDOQb6kR87wg8hV | 38,169 | https://github.com/openclaw/openclaw/issues/38169 | https://api.github.com/repos/openclaw/openclaw/issues/38169 | ugc-factory-01 node cannot connect: DNS resolution failure for Tailscale MagicDNS | ## Summary
The ugc-factory-01 node is unable to connect to the gateway due to DNS resolution failure.
## Symptoms
- Node service in restart loop (17,000+ restarts)
- Error: `getaddrinfo ENOTFOUND srv1325349.tail657eaf.ts.net`
- Tailscale shows host as **online** but OpenClaw node is **disconnected**
## Root Cause
The... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-06T16:50:36Z | 2026-04-27T05:01:59Z | 2026-04-24T19:31:04Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | TerminalGravity | 15,236,602 | MDQ6VXNlcjE1MjM2NjAy | User | false |
openclaw/openclaw | 4,035,305,520 | I_kwDOQb6kR87wheAw | 38,212 | https://github.com/openclaw/openclaw/issues/38212 | https://api.github.com/repos/openclaw/openclaw/issues/38212 | codex-cli/gpt-5.4 fails in embedded/helper paths while openai-codex/gpt-5.4 works | **Summary**
On OpenClaw `2026.3.2`, helper paths that use the embedded runner can still fail when the configured primary model is `codex-cli/gpt-5.4`, even though the equivalent Codex OAuth-backed ref `openai-codex/gpt-5.4` already has forward-compat support.
The most visible failure path is the bundled `session-memo... | open | null | false | 2 | [] | [] | 2026-03-06T17:24:21Z | 2026-05-04T06:03:46Z | null | NONE | null | 20260504T090105Z | 2026-05-04T09:01:05Z | huangyongde | 40,013,727 | MDQ6VXNlcjQwMDEzNzI3 | User | false |
openclaw/openclaw | 4,035,315,022 | I_kwDOQb6kR87whgVO | 38,217 | https://github.com/openclaw/openclaw/issues/38217 | https://api.github.com/repos/openclaw/openclaw/issues/38217 | [Bug] Feishu plugin config validation fails when appId/appSecret in wrong location | # 🐛 飞书插件配置校验失败导致 Gateway 无法启动
## Plugin Version
2026.3.7-beta.1
## OpenClaw Version
2026.3.2
## Bug Description
配置飞书插件 `feishu-openclaw-plugin` 时，如果将 `appId` 和 `appSecret` 直接放在 `plugins.entries.feishu-openclaw-plugin.config` 中，OpenClaw 会提示配置校验失败：
```
Invalid config at ~/.openclaw/openclaw.json:
- plugins.entrie... | closed | completed | true | 2 | [
"stale"
] | [] | 2026-03-06T17:26:14Z | 2026-04-27T05:01:58Z | 2026-04-24T19:31:16Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | RAFOLIE | 129,680,436 | U_kgDOB7rENA | User | false |
openclaw/openclaw | 4,035,343,876 | I_kwDOQb6kR87whnYE | 38,223 | https://github.com/openclaw/openclaw/issues/38223 | https://api.github.com/repos/openclaw/openclaw/issues/38223 | [Bug]: WebUI displays "No output" for all tool results even though stdout contains data | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
**Bug title:** WebUI displays "No output" for all tool results even though stdout contains data
**Bug Description:**
The OpenClaw WebUI (Control UI) shows that all tool executions result is "No output — tool completed successfully", but th... | closed | completed | true | 2 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-06T17:34:01Z | 2026-04-27T05:01:58Z | 2026-04-24T19:31:21Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | dazexcl | 8,541,447 | MDQ6VXNlcjg1NDE0NDc= | User | false |
openclaw/openclaw | 4,035,367,485 | I_kwDOQb6kR87whtI9 | 38,231 | https://github.com/openclaw/openclaw/issues/38231 | https://api.github.com/repos/openclaw/openclaw/issues/38231 | Feature Request: Restore recursive skill scanning for better skill organization | ### Description
Currently, OpenClaw (v2026.3.2) only seems to scan the first level of the `~/.openclaw/skills/` directory. This makes it difficult to organize skills, especially for 'bundle' repositories or multi-skill projects like `agent-browser-skills`, where individual skills are located in subdirectories (e.g., `a... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-06T17:40:24Z | 2026-04-28T04:43:39Z | 2026-04-26T01:33:58Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | zesty-clawd | 258,447,987 | U_kgDOD2eacw | User | false |
openclaw/openclaw | 4,035,390,949 | I_kwDOQb6kR87why3l | 38,235 | https://github.com/openclaw/openclaw/issues/38235 | https://api.github.com/repos/openclaw/openclaw/issues/38235 | [Slack] Agent identity (username/icon) not passed through streaming paths | ## Bug Description
When `streaming` is set to `true` or `partial`, Slack messages show the default Slack app name instead of the custom agent identity (name/emoji/icon). With `streaming: false`, the correct agent name and icon appear.
## Root Cause (from source analysis)
The non-streaming path correctly resolves `ou... | open | null | false | 3 | [] | [] | 2026-03-06T17:46:46Z | 2026-05-04T06:03:49Z | null | NONE | null | 20260504T090105Z | 2026-05-04T09:01:05Z | lacymorrow | 1,311,301 | MDQ6VXNlcjEzMTEzMDE= | User | false |
openclaw/openclaw | 4,035,400,190 | I_kwDOQb6kR87wh1H- | 38,236 | https://github.com/openclaw/openclaw/issues/38236 | https://api.github.com/repos/openclaw/openclaw/issues/38236 | Bug: Browser processes not cleaned up after automation tasks | # Bug: Browser processes not cleaned up after automation tasks
## Description
When using the browser automation tool, Chrome renderer processes remain running after tasks complete, causing:
- High CPU usage (accumulated to 1000%+ across multiple processes)
- Memory bloat (3.8GB+ from a single zombie process)
- System... | closed | completed | true | 2 | [
"stale"
] | [] | 2026-03-06T17:49:25Z | 2026-04-28T04:45:37Z | 2026-04-25T23:35:19Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | tianxiao1430-jpg | 255,869,001 | U_kgDOD0BASQ | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.