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 | 4,121,209,233 | I_kwDOQb6kR871pKmR | 52,914 | https://github.com/openclaw/openclaw/issues/52914 | https://api.github.com/repos/openclaw/openclaw/issues/52914 | [Bug]: WeChat QR code scan shows blank/white screen (QCLAW_USER_GUID and QCLAW_USER_ID not injected) | ### Bug type
Regression (worked before, now fails)
### Summary
QClaw Version: 0.1.16
Environment:
- PC: Windows 11 (x64)
- Phone: iOS, WeChat (latest version)
Description:
After scanning the WeChat connection QR code in QClaw settings, the WeChat app on iPhone shows a blank white screen and fails to load the page.... | closed | not_planned | true | 3 | [
"bug",
"regression"
] | [] | 2026-03-23T14:10:36Z | 2026-04-30T04:46:48Z | 2026-04-25T06:10:37Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | qlibj | 230,280,380 | U_kgDODbnMvA | User | false |
openclaw/openclaw | 4,121,374,909 | I_kwDOQb6kR871pzC9 | 52,929 | https://github.com/openclaw/openclaw/issues/52929 | https://api.github.com/repos/openclaw/openclaw/issues/52929 | Gateway fails to clean up stale .jsonl.lock file on hung/crashed session | ## Summary
The gateway process fails to clean up session lock files when a session hangs or the gateway is killed uncleanly. This causes subsequent gateway starts to get stuck indefinitely (shows as "typing..." in Discord) until the lock file is manually deleted.
## Steps to Reproduce
1. Let the gateway run for a wh... | closed | completed | true | 4 | [] | [] | 2026-03-23T14:35:34Z | 2026-04-30T04:46:47Z | 2026-04-25T06:10:44Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | toekneepark | 17,776,836 | MDQ6VXNlcjE3Nzc2ODM2 | User | false |
openclaw/openclaw | 4,121,764,876 | I_kwDOQb6kR871rSQM | 52,973 | https://github.com/openclaw/openclaw/issues/52973 | https://api.github.com/repos/openclaw/openclaw/issues/52973 | Bundled provider plugins can fail to resolve plugin-sdk/provider-entry from source | ## Summary
On current `main` (`2026.3.23`, commit `d2e8ed36`), bundled provider plugins loaded from source can fail to resolve `openclaw/plugin-sdk/provider-entry` and end up trying to load a synthetic path under `dist/plugin-sdk/root-alias.cjs/provider-entry`.
In my local runtime this shows up repeatedly for the bund... | closed | completed | true | 1 | [] | [] | 2026-03-23T15:31:18Z | 2026-04-30T04:46:45Z | 2026-04-25T06:11:03Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | UncleJ-h | 70,625,491 | MDQ6VXNlcjcwNjI1NDkx | User | false |
openclaw/openclaw | 4,121,765,033 | I_kwDOQb6kR871rSSp | 52,974 | https://github.com/openclaw/openclaw/issues/52974 | https://api.github.com/repos/openclaw/openclaw/issues/52974 | [RFC] memory-nexo: Cognitive memory backend with Atkinson-Shiffrin model | ## Summary
Proposing `memory-nexo` as a third memory backend for OpenClaw, alongside `memory-core` and `memory-lancedb`. This implements the `kind: "memory"` plugin slot with a cognitive architecture based on established psychology models.
**Note:** CONTRIBUTING.md recommends a Discussion for architectural proposals,... | closed | not_planned | true | 3 | [] | [] | 2026-03-23T15:31:19Z | 2026-04-30T04:46:44Z | 2026-04-25T06:11:10Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | wazionapps | 216,472,602 | U_kgDODOccGg | User | false |
openclaw/openclaw | 4,121,823,186 | I_kwDOQb6kR871rgfS | 52,981 | https://github.com/openclaw/openclaw/issues/52981 | https://api.github.com/repos/openclaw/openclaw/issues/52981 | Alibaba Model Studio: Context tokens show 0 (prompt_tokens not recognized) | ## Bug Description
When using Alibaba Model Studio models (e.g., `modelstudio/qwen3.5-plus`, `modelstudio/glm-5`), OpenClaw displays **0 context tokens used** even during active conversations. The context window size shows correctly, but "used" is always 0.
## Root Cause
In `dist/pi-embedded-CzQCqSlH.js` (line ~1813... | closed | completed | true | 3 | [] | [] | 2026-03-23T15:40:52Z | 2026-04-30T04:46:43Z | 2026-04-25T06:11:17Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | FiredMosquito831 | 79,049,343 | MDQ6VXNlcjc5MDQ5MzQz | User | false |
openclaw/openclaw | 4,121,984,617 | I_kwDOQb6kR871sH5p | 52,994 | https://github.com/openclaw/openclaw/issues/52994 | https://api.github.com/repos/openclaw/openclaw/issues/52994 | WhatsApp plugin (0.0.5-Alpha) fails to install — missing openclaw.extensions in package.json | ## Bug Description
The WhatsApp plugin (`whatsapp@0.0.5-Alpha`) downloads successfully from ClawHub but fails to install because its `package.json` is missing required fields.
## Steps to Reproduce
```bash
openclaw plugins install whatsapp@0.0.5-Alpha
```
## Error Output
```
Resolving clawhub:whatsapp@0.0.5-Alpha…... | closed | completed | true | 1 | [] | [] | 2026-03-23T16:05:53Z | 2026-04-30T04:46:42Z | 2026-04-25T06:11:23Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | bubucilo | 262,674,519 | U_kgDOD6gYVw | User | false |
openclaw/openclaw | 4,122,140,209 | I_kwDOQb6kR871st4x | 53,002 | https://github.com/openclaw/openclaw/issues/53002 | https://api.github.com/repos/openclaw/openclaw/issues/53002 | [Bug]: local TS extensions importing openclaw/plugin-sdk/plugin-entry fail to load on 2026.3.22 | ### Bug type
Regression / plugin loading
### Summary
On OpenClaw 2026.3.22, a local extension under `~/.openclaw/extensions/<id>/index.ts` that imports `openclaw/plugin-sdk/plugin-entry` fails to load because the plugin loader resolves the import to `.../dist/plugin-sdk/root-alias.cjs/plugin-entry` instead of the scop... | closed | completed | true | 4 | [] | [] | 2026-03-23T16:30:51Z | 2026-04-30T04:46:41Z | 2026-04-25T06:11:36Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | senshinya | 155,552,943 | U_kgDOCUWMrw | User | false |
openclaw/openclaw | 4,122,158,032 | I_kwDOQb6kR871syPQ | 53,003 | https://github.com/openclaw/openclaw/issues/53003 | https://api.github.com/repos/openclaw/openclaw/issues/53003 | openclaw-lark v2026.3.18 incompatible with OpenClaw v2026.3.22 (plugin-sdk breaking changes) | ## Problem Description
After upgrading OpenClaw to the latest version (2026.3.22), the Feishu (Lark) plugin stops working.
## Root Cause
OpenClaw 2026.3.22 introduced a **major breaking refactor** of the plugin-sdk:
- Many old APIs were removed (e.g. \`resolveSenderCommandAuthorization\`, \`buildChannelConfigSchema... | closed | not_planned | true | 3 | [] | [] | 2026-03-23T16:33:14Z | 2026-04-30T04:46:41Z | 2026-04-25T06:11:43Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | 0xcjl | 38,884,292 | MDQ6VXNlcjM4ODg0Mjky | User | false |
openclaw/openclaw | 4,122,228,930 | I_kwDOQb6kR871tDjC | 53,012 | https://github.com/openclaw/openclaw/issues/53012 | https://api.github.com/repos/openclaw/openclaw/issues/53012 | Control UI: show session displayName in chat dropdown instead of raw key | ## Problem
The chat session dropdown in the Control UI shows the raw session key (e.g. `agent:main:telegram:group:-5034552618`), which is not human-readable — especially for group chats.
## Current behavior
The gateway already resolves a `displayName` for sessions (visible via `sessions.list`), e.g. `telegram:g-open... | closed | completed | true | 1 | [] | [] | 2026-03-23T16:45:18Z | 2026-04-30T04:46:39Z | 2026-04-25T06:11:56Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | nachtsheim | 2,680,978 | MDQ6VXNlcjI2ODA5Nzg= | User | false |
openclaw/openclaw | 4,122,327,900 | I_kwDOQb6kR871tbtc | 53,022 | https://github.com/openclaw/openclaw/issues/53022 | https://api.github.com/repos/openclaw/openclaw/issues/53022 | [Community Contribution]: Local Intent Firewall, HAM Memory, & Dynamic Model Routing (OC-Argent) | ### Summary
I have been running OpenClaw in a production environment and hit significant friction with the default flat-file memory structure, escalating API costs, and a lack of tool-execution oversight.
To solve this natively without relying on external cloud databases, I built a drop-in architectural template ins... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-23T17:02:14Z | 2026-04-30T04:46:38Z | 2026-04-25T06:12:02Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Ravenspo | 261,277,261 | U_kgDOD5LGTQ | User | false |
openclaw/openclaw | 4,122,330,984 | I_kwDOQb6kR871tcdo | 53,025 | https://github.com/openclaw/openclaw/issues/53025 | https://api.github.com/repos/openclaw/openclaw/issues/53025 | Feature request: multi-instance OpenClaw collaboration with shared artifact handoff | ### Summary
**Enable collaboration between multiple OpenClaw instances using private workspaces and shared artifact-based handoff, instead of relying solely on text-based agent communication.**
### Problem to solve
OpenClaw’s current multi-agent model works well within a single instance, where agents share context a... | closed | not_planned | true | 7 | [
"enhancement"
] | [] | 2026-03-23T17:02:49Z | 2026-04-30T04:46:37Z | 2026-04-25T06:12:09Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | nanami-he | 270,413,913 | U_kgDOEB4wWQ | User | false |
openclaw/openclaw | 4,067,523,684 | I_kwDOQb6kR87ycXxk | 44,402 | https://github.com/openclaw/openclaw/issues/44402 | https://api.github.com/repos/openclaw/openclaw/issues/44402 | [Bug]: `openclaw gateway install` does not capture Nix (Home Manager) PATH entries in LaunchAgent plist | ## Bug description
`openclaw gateway install` generates a LaunchAgent plist (`~/Library/LaunchAgents/ai.openclaw.gateway.plist`) that captures the current shell PATH into the `EnvironmentVariables` dict. However, paths added by Nix Home Manager (e.g. `~/.nix-profile/bin`) are **not included** in the generated plist, e... | closed | completed | true | 5 | [] | [] | 2026-03-12T22:00:48Z | 2026-04-30T04:46:37Z | 2026-04-25T06:12:12Z | CONTRIBUTOR | null | 20260430T060018Z | 2026-04-30T06:00:18Z | jerome-benoit | 6,537,332 | MDQ6VXNlcjY1MzczMzI= | User | false |
openclaw/openclaw | 4,122,331,391 | I_kwDOQb6kR871tcj_ | 53,026 | https://github.com/openclaw/openclaw/issues/53026 | https://api.github.com/repos/openclaw/openclaw/issues/53026 | ui:build installs production-only deps and breaks Control UI asset generation | `pnpm ui:build` currently auto-runs `pnpm install --prod` when UI dependencies are missing. That strips `devDependencies`, so `vite` is unavailable and the build fails before `dist/control-ui/index.html` is generated.
Repro:
1. Remove `ui/node_modules` or start from a fresh checkout.
2. Run `pnpm ui:build`.
3. The scr... | closed | completed | true | 1 | [] | [] | 2026-03-23T17:02:54Z | 2026-04-30T04:46:36Z | 2026-04-25T06:12:15Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | apoapostolov | 4,083,812 | MDQ6VXNlcjQwODM4MTI= | User | false |
openclaw/openclaw | 4,122,379,499 | I_kwDOQb6kR871toTr | 53,033 | https://github.com/openclaw/openclaw/issues/53033 | https://api.github.com/repos/openclaw/openclaw/issues/53033 | Bug: Control UI 2026.3.22 - Vite build produces invalid JS syntax + CSP too strict | ## Bug Description
OpenClaw 2026.3.22 Control UI has critical issues preventing the dashboard from loading:
### Issue 1: Vite Build Produces Invalid JavaScript
- **File**: `dist/control-ui/assets/index-DPRTqWHM.js`
- **Error**: `Uncaught SyntaxError: Invalid or unexpected token (at line 813)`
- **Location**: Minified... | closed | completed | true | 1 | [] | [] | 2026-03-23T17:12:23Z | 2026-04-30T04:46:35Z | 2026-04-25T06:12:22Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | imHuangke | 229,515,547 | U_kgDODa4hGw | User | false |
openclaw/openclaw | 4,122,478,075 | I_kwDOQb6kR871uAX7 | 53,048 | https://github.com/openclaw/openclaw/issues/53048 | https://api.github.com/repos/openclaw/openclaw/issues/53048 | 43 bundled plugins show 'plugin id mismatch' warnings in v2026.3.22 | ## Description
After upgrading to v2026.3.22 (stable, git install), all 43 bundled plugins emit `plugin id mismatch` warnings on every `openclaw doctor`, `openclaw cron list`, `openclaw update --dry-run`, and other CLI commands.
## Example
```
plugins.entries.openrouter: plugin openrouter: plugin id mismatch (manife... | closed | completed | true | 2 | [] | [] | 2026-03-23T17:30:23Z | 2026-04-30T04:46:34Z | 2026-04-25T06:12:28Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | ldfandian | 5,018,331 | MDQ6VXNlcjUwMTgzMzE= | User | false |
openclaw/openclaw | 4,122,732,175 | I_kwDOQb6kR871u-aP | 53,070 | https://github.com/openclaw/openclaw/issues/53070 | https://api.github.com/repos/openclaw/openclaw/issues/53070 | 2026.3.22: breaking config schema removals + describeMessageTool TypeError silently break Discord tool calling | ## Summary
Two regressions introduced in **2026.3.22** that together silently break agent tool calling.
---
## Issue 1: Removed config keys produce hard errors with no migration path
**Version:** 2026.3.22
**Severity:** High — blocks gateway startup with `Invalid config` and disrupts tool registration
... | closed | completed | true | 1 | [] | [] | 2026-03-23T18:13:49Z | 2026-04-30T04:46:34Z | 2026-04-25T06:12:35Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | ZachariBarnes | 10,678,967 | MDQ6VXNlcjEwNjc4OTY3 | User | false |
openclaw/openclaw | 4,122,732,795 | I_kwDOQb6kR871u-j7 | 53,071 | https://github.com/openclaw/openclaw/issues/53071 | https://api.github.com/repos/openclaw/openclaw/issues/53071 | 2026.3.22: breaking config schema removals + describeMessageTool TypeError silently break Discord tool calling | ## Summary
Two regressions introduced in **2026.3.22** that together silently break agent tool calling.
---
## Issue 1: Removed config keys produce hard errors with no migration path
**Version:** 2026.3.22
**Severity:** High — blocks gateway startup with `Invalid config` and disrupts tool registration
... | closed | completed | true | 2 | [] | [] | 2026-03-23T18:13:53Z | 2026-04-30T04:46:33Z | 2026-04-25T06:12:41Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | ZachariBarnes | 10,678,967 | MDQ6VXNlcjEwNjc4OTY3 | User | false |
openclaw/openclaw | 4,122,845,171 | I_kwDOQb6kR871vZ_z | 53,076 | https://github.com/openclaw/openclaw/issues/53076 | https://api.github.com/repos/openclaw/openclaw/issues/53076 | Priority lanes: interactive messages should bypass queued system events | ## Problem
When multiple system events (diff-triggered heartbeats, exec failure notifications, stall recovery dispatches) pile into the main agent session queue simultaneously, interactive user messages (e.g. Telegram DMs) get blocked behind them. Observed a 5-minute wait (`waitedMs=299279`) for a user message because... | closed | completed | true | 2 | [] | [] | 2026-03-23T18:30:50Z | 2026-04-30T04:46:32Z | 2026-04-25T06:12:48Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | CodexOperator | 172,000,056 | U_kgDOCkCDOA | User | false |
openclaw/openclaw | 4,123,031,191 | I_kwDOQb6kR871wHaX | 53,090 | https://github.com/openclaw/openclaw/issues/53090 | https://api.github.com/repos/openclaw/openclaw/issues/53090 | Paperclip openclaw_gateway agents: WebSocket closed immediately even with auth=none | ## Environment
- OpenClaw: 2026.3.13
- Paperclip: latest (npx paperclipai)
- Server: Ubuntu 24.04, ARM64 (Hetzner CAX11)
- Gateway: loopback binding, token auth
## Problem
Paperclip agents configured with `adapterType: "openclaw_gateway"` cannot connect to the OpenClaw gateway. The WebSocket connection is closed imme... | closed | completed | true | 1 | [] | [] | 2026-03-23T18:58:00Z | 2026-04-30T04:46:31Z | 2026-04-25T06:12:54Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | eburfield-second | 261,271,902 | U_kgDOD5KxXg | User | false |
openclaw/openclaw | 4,123,113,193 | I_kwDOQb6kR871wbbp | 53,096 | https://github.com/openclaw/openclaw/issues/53096 | https://api.github.com/repos/openclaw/openclaw/issues/53096 | [Bug]: Control UI assets not found. Build them with `pnpm ui:build` (auto-installs UI deps), or run `pnpm ui:dev` during development. | ### Bug type
Regression (worked before, now fails)
### Summary
2026.3.22 - Did an Update Now from web browser interface to 2026.3.22 and now browser ui doesnt work. With error in subject. Same as #4855 though that is marked closed!?
### Steps to reproduce
1. Update now from browser interface to 2026.3.22 on OSX
... | closed | completed | true | 13 | [
"bug",
"regression"
] | [] | 2026-03-23T19:11:56Z | 2026-04-30T04:46:30Z | 2026-04-25T06:13:01Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | rajeshkrishnamurthy | 17,078,943 | MDQ6VXNlcjE3MDc4OTQz | User | false |
openclaw/openclaw | 4,123,131,748 | I_kwDOQb6kR871wf9k | 53,097 | https://github.com/openclaw/openclaw/issues/53097 | https://api.github.com/repos/openclaw/openclaw/issues/53097 | Feature request: /v1/agents endpoint for agent discovery | ## Use Case
Third-party apps (like Pane) that integrate with OpenClaw need to discover which agents are configured, their roles, models, and workspace paths. Currently this information only exists in `openclaw.json` on disk — there's no API to query it.
## Proposed Endpoint
```
GET /v1/agents
Response:
{
"agents"... | closed | completed | true | 2 | [] | [] | 2026-03-23T19:15:18Z | 2026-04-28T04:59:23Z | 2026-04-25T06:13:07Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | matthewcarano | 30,149,603 | MDQ6VXNlcjMwMTQ5NjAz | User | false |
openclaw/openclaw | 4,123,293,615 | I_kwDOQb6kR871xHev | 53,109 | https://github.com/openclaw/openclaw/issues/53109 | https://api.github.com/repos/openclaw/openclaw/issues/53109 | Better handling for interrupted streaming tool calls to avoid silent failure and retry loops | ## Summary
When an upstream streaming response is interrupted during a long tool call, OpenClaw can end up in a misleading state where the requested work was not completed, but the session does not fail clearly enough for the user or the next turn.
In practice, this can look like:
- the assistant appears to have prog... | closed | completed | true | 1 | [] | [] | 2026-03-23T19:45:13Z | 2026-04-30T04:46:30Z | 2026-04-25T06:13:14Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Schnitzel | 750,999 | MDQ6VXNlcjc1MDk5OQ== | User | false |
openclaw/openclaw | 4,123,333,624 | I_kwDOQb6kR871xRP4 | 53,113 | https://github.com/openclaw/openclaw/issues/53113 | https://api.github.com/repos/openclaw/openclaw/issues/53113 | WhatsApp: inbound messages silently lost after reconnect (append-type messages dropped) — production-tested fix included | ## Problem
WhatsApp messages sent to the bot during a disconnection (408 timeout, Bad MAC burst, prekey renegotiation) are silently dropped. The messages arrive via Baileys as `upsert.type === "append"` during reconnection sync, but OpenClaw unconditionally skips them with:
```js
if (upsert.type === "append") continu... | closed | completed | true | 2 | [] | [] | 2026-03-23T19:52:16Z | 2026-04-30T04:46:29Z | 2026-04-25T06:13:20Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | marieldejesus12 | 38,159,708 | MDQ6VXNlcjM4MTU5NzA4 | User | false |
openclaw/openclaw | 4,123,413,678 | I_kwDOQb6kR871xkyu | 53,118 | https://github.com/openclaw/openclaw/issues/53118 | https://api.github.com/repos/openclaw/openclaw/issues/53118 | [Feature]: Fix multi messages prompts | ### Summary
Openclaw can't understand correctly if I send a request with 2 messages in Telegram. Sometimes I write and send the message without thinking then realize it needs more context and add another one. But the first one is already taken to process.
It's too complex to call /stop then resend second message ag... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-03-23T20:08:31Z | 2026-04-30T04:46:28Z | 2026-04-25T06:13:27Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | asidko | 22,843,881 | MDQ6VXNlcjIyODQzODgx | User | false |
openclaw/openclaw | 4,123,425,108 | I_kwDOQb6kR871xnlU | 53,120 | https://github.com/openclaw/openclaw/issues/53120 | https://api.github.com/repos/openclaw/openclaw/issues/53120 | Approval flow regression: '/approve <id> allow-always' frequently returns 'unknown or expired approval id' for node system.run denies | ## Summary
After updating to OpenClaw 2026.3.8/2026.3.13-era builds, node `system.run` deny approvals are not reliably consumable from Control UI. By the time `/approve <id> allow-always` is submitted, gateway frequently returns:
`GatewayClientRequestError: unknown or expired approval id`
This blocks recovery from `a... | closed | completed | true | 2 | [] | [] | 2026-03-23T20:10:45Z | 2026-04-30T04:46:27Z | 2026-04-25T06:13:34Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | dabingougou | 6,183,426 | MDQ6VXNlcjYxODM0MjY= | User | false |
openclaw/openclaw | 4,123,443,972 | I_kwDOQb6kR871xsME | 53,122 | https://github.com/openclaw/openclaw/issues/53122 | https://api.github.com/repos/openclaw/openclaw/issues/53122 | [Feature]: Expose onToolStart/onCompactionStart callbacks to plugin-sdk | ### Summary
Allow channel plugins to react to agent lifecycle events (tool calls, compaction) by exposing onToolStart/onCompactionStart callbacks through dispatchInboundReplyWithBase.
### Problem to solve
Channel plugins (Max Messenger, WhatsApp, etc.) cannot react to agent lifecycle events. When the agent works wit... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-03-23T20:13:43Z | 2026-04-30T04:46:26Z | 2026-04-25T06:13:40Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | jeket3 | 145,283,156 | U_kgDOCKjYVA | User | false |
openclaw/openclaw | 4,123,481,631 | I_kwDOQb6kR871x1Yf | 53,124 | https://github.com/openclaw/openclaw/issues/53124 | https://api.github.com/repos/openclaw/openclaw/issues/53124 | Telegram streaming: preview deleted before media message confirmed sent (group chat) | ## Summary
When `channels.telegram.streaming` is set to `partial` (default since 2026.3.22), agent responses containing media (e.g. images) in **group chats** result in a "deleted message" with no media delivered.
The root cause: when `hasMedia=true`, `canEditViaPreview=false` — the response is sent via `sendPayload`... | closed | completed | true | 4 | [] | [] | 2026-03-23T20:20:12Z | 2026-04-30T04:46:26Z | 2026-04-25T06:13:47Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | mvalentsev | 2,305,718 | MDQ6VXNlcjIzMDU3MTg= | User | false |
openclaw/openclaw | 4,123,500,697 | I_kwDOQb6kR871x6CZ | 53,128 | https://github.com/openclaw/openclaw/issues/53128 | https://api.github.com/repos/openclaw/openclaw/issues/53128 | [Bug]: `onboard --install-daemon` does not set `gateway.remote.token`, causing immediate token mismatch | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Running `openclaw onboard --install-daemon` on a fresh install sets `gateway.auth.token` but does not set `gateway.remote.token`. This causes an immediate token mismatch error when the CLI tries to connect to the gateway it just started:
`... | closed | not_planned | true | 2 | [] | [] | 2026-03-23T20:23:26Z | 2026-04-30T04:46:25Z | 2026-04-25T06:13:53Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | pmatos | 7,911 | MDQ6VXNlcjc5MTE= | User | false |
openclaw/openclaw | 4,123,602,484 | I_kwDOQb6kR871yS40 | 53,141 | https://github.com/openclaw/openclaw/issues/53141 | https://api.github.com/repos/openclaw/openclaw/issues/53141 | [Bug]: Telegram ElevenLabs TTS voices short/medium replies but skips long replies (text-only) | ## Summary
In Telegram DM testing with ElevenLabs TTS working, short and medium replies are delivered as voiced responses, but long-form replies fall back to text-only with no voice output.
## Environment
- Channel: Telegram DM
- TTS provider: ElevenLabs
- Voice output confirmed working for short/medium replies
- Long... | closed | completed | true | 2 | [] | [] | 2026-03-23T20:41:44Z | 2026-04-30T04:46:21Z | 2026-04-25T06:14:20Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | A1fred-AI | 269,621,014 | U_kgDOEBIXFg | User | false |
openclaw/openclaw | 4,123,613,855 | I_kwDOQb6kR871yVqf | 53,143 | https://github.com/openclaw/openclaw/issues/53143 | https://api.github.com/repos/openclaw/openclaw/issues/53143 | Config normalization at startup drops Slack channel startup task (cdd797f) | ## Bug Description
After upgrading from `d41c9ad` to `cdd797f` (2026.3.23-beta.1), the gateway's Slack Socket Mode channel never starts. The gateway completes plugin loading and reaches `listening on ws://0.0.0.0:18789`, but the Slack channel `startAccount` / `app.start()` phase is never entered.
## Root Cause (narro... | closed | completed | true | 4 | [] | [] | 2026-03-23T20:43:48Z | 2026-04-30T04:46:21Z | 2026-04-25T06:14:27Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | chrisbaker2000 | 174,035,702 | U_kgDOCl-S9g | User | false |
openclaw/openclaw | 4,123,677,124 | I_kwDOQb6kR871ylHE | 53,150 | https://github.com/openclaw/openclaw/issues/53150 | https://api.github.com/repos/openclaw/openclaw/issues/53150 | [Feature Request] Support dynamic model switching in Gateway chat completions API | # OpenClaw Issue - 多模型切换功能请求
## 标题
[Feature Request] Support dynamic model switching in Gateway chat completions API
## 正文
### Problem
Currently, the Gateway's `/v1/chat/completions` endpoint ignores the `model` parameter in the request body and always uses the agent's default model.
### Current Behavior
When sendi... | closed | completed | true | 1 | [] | [] | 2026-03-23T20:55:07Z | 2026-04-30T04:46:19Z | 2026-04-25T06:14:34Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | kumikoda110 | 15,316,691 | MDQ6VXNlcjE1MzE2Njkx | User | false |
openclaw/openclaw | 4,123,760,147 | I_kwDOQb6kR871y5YT | 53,158 | https://github.com/openclaw/openclaw/issues/53158 | https://api.github.com/repos/openclaw/openclaw/issues/53158 | Session keys use ':openai:' regardless of actual LLM provider | ## Problem
Subagent session keys are generated with `:openai:` in the key string regardless of the actual model/provider being used.
**Example:**
```
agent:main:openai:30a4c31c-5d64-46f7-8189-75b439d80264
```
This session was running `claude-opus-4-6` (Anthropic), not OpenAI.
## Scope
In my instance, 1,492 out of 1... | closed | completed | true | 1 | [] | [] | 2026-03-23T21:11:02Z | 2026-04-30T04:46:18Z | 2026-04-25T06:14:40Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | coolmanns | 117,685,070 | U_kgDOBwO7Tg | User | false |
openclaw/openclaw | 4,123,801,379 | I_kwDOQb6kR871zDcj | 53,163 | https://github.com/openclaw/openclaw/issues/53163 | https://api.github.com/repos/openclaw/openclaw/issues/53163 | [Bug] message tool file attachments silently dropped on Discord after v2026.3.22 | ## Description
After updating from v2026.3.13 → v2026.3.22, the `message` tool's `filePath`, `media`, and `filename` parameters silently drop file attachments on Discord. The tool returns `{ok: true}` with a valid `messageId`, but the message arrives with `"attachments": []` — no file is uploaded.
## Steps to Reprodu... | closed | completed | true | 2 | [] | [] | 2026-03-23T21:19:25Z | 2026-04-30T04:46:18Z | 2026-04-25T06:14:47Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | LAG-Yadav | 66,820,006 | MDQ6VXNlcjY2ODIwMDA2 | User | false |
openclaw/openclaw | 4,123,804,631 | I_kwDOQb6kR871zEPX | 53,164 | https://github.com/openclaw/openclaw/issues/53164 | https://api.github.com/repos/openclaw/openclaw/issues/53164 | ClawHub security scan false positive: cc-soul skill flagged as suspicious | ## Summary
Our skill **cc-soul** (slug: `cc-soul`, author: `wenroudeyu-collab`) is consistently flagged as "Suspicious - Medium Confidence" by the ClawHub security scanner, despite 0/65 VirusTotal detections.
## The issue
The scanner flags cc-soul because:
1. SKILL.md contains no inline code — it instructs users to ... | closed | not_planned | true | 1 | [] | [] | 2026-03-23T21:20:11Z | 2026-04-30T04:46:17Z | 2026-04-25T06:14:54Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | wenroudeyu-collab | 263,342,295 | U_kgDOD7JI1w | User | false |
openclaw/openclaw | 4,123,830,796 | I_kwDOQb6kR871zKoM | 53,167 | https://github.com/openclaw/openclaw/issues/53167 | https://api.github.com/repos/openclaw/openclaw/issues/53167 | Session lane lock not released after API error (529 overload) | # Session lane lock not released after API error (529 overload)
## Bug Description
When an Anthropic API call returns a 529 (service overloaded) error, the session run ends with `isError=true` but the session lane lock is never released. This causes the session to become permanently unresponsive to new inbound messag... | closed | completed | true | 3 | [] | [] | 2026-03-23T21:26:38Z | 2026-04-30T04:46:15Z | 2026-04-25T06:15:00Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | esulcj | 11,564,095 | MDQ6VXNlcjExNTY0MDk1 | User | false |
openclaw/openclaw | 4,123,833,702 | I_kwDOQb6kR871zLVm | 53,170 | https://github.com/openclaw/openclaw/issues/53170 | https://api.github.com/repos/openclaw/openclaw/issues/53170 | Feature request: BM25 hybrid search for memory_search | ## Problem
Currently `memory_search` uses embedding-only retrieval (text-embedding-3-large). In production across a multi-agent fleet, we're seeing a **48.5% empty retrieval rate** — nearly half of all memory searches return zero results due to vocabulary mismatch between query terms and stored content.
Additionally,... | closed | completed | true | 1 | [] | [] | 2026-03-23T21:27:28Z | 2026-04-30T04:46:15Z | 2026-04-25T06:15:07Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | landenlucas33 | 261,846,925 | U_kgDOD5t3jQ | User | false |
openclaw/openclaw | 4,123,857,708 | I_kwDOQb6kR871zRMs | 53,172 | https://github.com/openclaw/openclaw/issues/53172 | https://api.github.com/repos/openclaw/openclaw/issues/53172 | Control UI: Logo image broken on dashboard/connect page | ## Description
The logo image on the Control UI connect/dashboard page appears broken (fails to load). This is visible when first navigating to the gateway dashboard at `/openclaw` before authenticating.
## Steps to Reproduce
1. Start the gateway: `openclaw gateway run`
2. Open the Control UI in a browser: `http://1... | closed | completed | true | 3 | [] | [] | 2026-03-23T21:34:16Z | 2026-04-30T04:46:14Z | 2026-04-25T06:15:13Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | jaredhill4 | 7,476,478 | MDQ6VXNlcjc0NzY0Nzg= | User | false |
openclaw/openclaw | 4,123,881,813 | I_kwDOQb6kR871zXFV | 53,176 | https://github.com/openclaw/openclaw/issues/53176 | https://api.github.com/repos/openclaw/openclaw/issues/53176 | [Feature]: Be able to see the LLM temperature and set it custom if needed. | ### Summary
I cannot find where I can see the current temperature set for agents and how to change that.
### Problem to solve
My agent's soul and identity and skills all have strict guidelines on how to act. But my agent keeps ignoring the settings. I have tried GLM 4.7, GLM 5.0, and now minimax, and even though th... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-03-23T21:41:09Z | 2026-04-30T04:46:13Z | 2026-04-25T06:15:20Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | najef1979-code | 233,555,716 | U_kgDODevHBA | User | false |
openclaw/openclaw | 4,123,969,520 | I_kwDOQb6kR871zsfw | 53,189 | https://github.com/openclaw/openclaw/issues/53189 | https://api.github.com/repos/openclaw/openclaw/issues/53189 | /status slash command shows 'Done' without status card after 3.13→3.22 upgrade | ## Description
After upgrading from OpenClaw v2026.3.13 to v2026.3.22, the `/status` slash command in Discord shows only **"✅ Done."** as an ephemeral response without rendering the status card (model, context usage, session info, etc.).
## Steps to Reproduce
1. Upgrade from v2026.3.13 to v2026.3.22 via `npm update ... | closed | completed | true | 7 | [] | [] | 2026-03-23T22:03:18Z | 2026-04-30T04:46:12Z | 2026-04-25T06:15:27Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | SwivelLabs | 261,235,255 | U_kgDOD5IiNw | User | false |
openclaw/openclaw | 4,124,000,829 | I_kwDOQb6kR871z0I9 | 53,192 | https://github.com/openclaw/openclaw/issues/53192 | https://api.github.com/repos/openclaw/openclaw/issues/53192 | Feature request: Native playbook support — reusable multi-agent pipeline specs in AGENTS.md / openclaw.json | ## Summary
We have been running a multi-agent workspace (11 specialist agents) on OpenClaw for several months and have independently converged on a pattern we are calling **playbooks** — reusable, parameterised multi-agent pipeline specs that standardise how complex recurring workflows are executed. We would like to p... | closed | not_planned | true | 1 | [] | [] | 2026-03-23T22:11:04Z | 2026-04-30T04:46:11Z | 2026-04-25T06:15:33Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | reddinft | 263,792,046 | U_kgDOD7klrg | User | false |
openclaw/openclaw | 4,124,005,040 | I_kwDOQb6kR871z1Kw | 53,194 | https://github.com/openclaw/openclaw/issues/53194 | https://api.github.com/repos/openclaw/openclaw/issues/53194 | [Feature]: Agent Feature Observation & Submission System | ### Summary
What it does:
Allow an agent to be designated, on a schedule or for a time window, specifically to observe gaps, friction points, and workarounds in their daily work or think of a feature that will add real value to Openclaw. The agent then submits structured feature ideas to a shared online queue where ot... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-23T22:12:08Z | 2026-04-30T04:46:10Z | 2026-04-25T06:15:40Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | najef1979-code | 233,555,716 | U_kgDODevHBA | User | false |
openclaw/openclaw | 4,124,012,015 | I_kwDOQb6kR871z23v | 53,195 | https://github.com/openclaw/openclaw/issues/53195 | https://api.github.com/repos/openclaw/openclaw/issues/53195 | Workspace bootstrap files not re-injected on session resume after gateway restart | ## Summary
Workspace bootstrap files (SOUL.md, TOOLS.md, IDENTITY.md, USER.md, AGENTS.md, etc.) are injected into session context at **session creation time** but are never re-injected when a session is **resumed** after a gateway restart.
This means edits to bootstrap files don't take effect in existing sessions unt... | closed | completed | true | 2 | [] | [] | 2026-03-23T22:14:07Z | 2026-04-30T04:46:10Z | 2026-04-25T06:15:46Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Stache73 | 7,196,384 | MDQ6VXNlcjcxOTYzODQ= | User | false |
openclaw/openclaw | 4,124,089,938 | I_kwDOQb6kR8710J5S | 53,201 | https://github.com/openclaw/openclaw/issues/53201 | https://api.github.com/repos/openclaw/openclaw/issues/53201 | [Bug]: Cron delivery misses post-subagent-announce output — only original run payloads are delivered | ## Bug type
Behavior bug (incorrect output/state without crash)
## Summary
When a cron isolated session spawns a subagent via `sessions_spawn`, the cron delivery mechanism only captures payloads from the **original** `runEmbeddedPiAgent` call. After the subagent completes and triggers the announce flow, the parent s... | closed | completed | true | 2 | [] | [] | 2026-03-23T22:33:15Z | 2026-04-30T04:46:09Z | 2026-04-25T06:15:53Z | CONTRIBUTOR | null | 20260430T060018Z | 2026-04-30T06:00:18Z | kagura-agent | 268,167,063 | U_kgDOD_vnlw | User | false |
openclaw/openclaw | 4,124,091,425 | I_kwDOQb6kR8710KQh | 53,202 | https://github.com/openclaw/openclaw/issues/53202 | https://api.github.com/repos/openclaw/openclaw/issues/53202 | [Bug]: Subagent announce hits repeated gateway timeouts (90s) in cron isolated sessions | ## Bug type
Behavior bug (timeout/retry without crash)
## Summary
When a cron isolated session spawns a subagent, the subagent completion announce flow hits repeated gateway timeouts (90 seconds each), causing retries and significant delays. In the observed case, retries 2/4 and 3/4 were logged, adding ~3+ minutes o... | closed | completed | true | 2 | [] | [] | 2026-03-23T22:33:34Z | 2026-04-30T04:46:08Z | 2026-04-25T06:15:59Z | CONTRIBUTOR | null | 20260430T060018Z | 2026-04-30T06:00:18Z | kagura-agent | 268,167,063 | U_kgDOD_vnlw | User | false |
openclaw/openclaw | 4,124,119,739 | I_kwDOQb6kR8710RK7 | 53,204 | https://github.com/openclaw/openclaw/issues/53204 | https://api.github.com/repos/openclaw/openclaw/issues/53204 | sessions_send announce retry blocks agent session for ~6 minutes on channel errors | ## Problem
When an agent uses `sessions_send` with `timeoutSeconds > 0` to query another agent, and the announce step fails (e.g. after a gateway restart when channels are temporarily unavailable), the **announce retry loop blocks the entire agent session** for several minutes.
### Observed behavior
1. Agent A calls... | closed | completed | true | 4 | [] | [] | 2026-03-23T22:40:32Z | 2026-04-30T04:46:07Z | 2026-04-25T06:16:06Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | claudiobedino | 259,268,385 | U_kgDOD3QfIQ | User | false |
openclaw/openclaw | 4,124,192,437 | I_kwDOQb6kR8710i61 | 53,215 | https://github.com/openclaw/openclaw/issues/53215 | https://api.github.com/repos/openclaw/openclaw/issues/53215 | [Feature]: OpenClaw MCP Server — Expose OpenClaw tools as Model Context Protocol tools for Claude/Cursor | ## 🎯 Problem Statement
The Model Context Protocol (MCP) ecosystem has exploded in 2025-2026, with top MCP servers accumulating 80k+ stars (awesome-mcp-servers: 83.9k⭐, playwright-mcp: 29.5k⭐, github-mcp-server: 28.2k⭐). MCP has become the de facto standard for AI agent tool integration, enabling Claude, Cursor, and o... | closed | completed | true | 1 | [] | [] | 2026-03-23T23:01:39Z | 2026-04-30T04:46:06Z | 2026-04-25T06:16:18Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | sykp241095 | 1,237,528 | MDQ6VXNlcjEyMzc1Mjg= | User | false |
openclaw/openclaw | 4,124,204,377 | I_kwDOQb6kR8710l1Z | 53,222 | https://github.com/openclaw/openclaw/issues/53222 | https://api.github.com/repos/openclaw/openclaw/issues/53222 | Inject session/channel context env vars into agent shell environments | ## Problem
When an OpenClaw agent spawns a tmux session (or any child process), there is no way to automatically know which OC session, channel, or agent originated the work. The shell environment has no context beyond `OPENCLAW_GATEWAY_PORT` and similar service-level vars.
This makes it impossible to auto-attribute ... | closed | completed | true | 1 | [] | [] | 2026-03-23T23:05:56Z | 2026-04-30T04:46:05Z | 2026-04-25T06:16:26Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | mbilenkis1 | 50,536,243 | MDQ6VXNlcjUwNTM2MjQz | User | false |
openclaw/openclaw | 4,124,206,589 | I_kwDOQb6kR8710mX9 | 53,223 | https://github.com/openclaw/openclaw/issues/53223 | https://api.github.com/repos/openclaw/openclaw/issues/53223 | WhatsApp bindings not routing to agent (Felipe) | ## Problem
WhatsApp bindings are not routing messages to the configured agent (Felipe). Messages arrive at the WhatsApp bridge but are not being delivered to the agent for processing.
## Expected Behavior
When a WhatsApp message arrives, it should be routed to the agent configured in the binding (e.g., Felipe agent ... | closed | not_planned | true | 12 | [] | [] | 2026-03-23T23:06:44Z | 2026-04-30T04:46:04Z | 2026-04-25T06:16:32Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | rafaelmariano-glitch | 259,843,811 | U_kgDOD3zm4w | User | false |
openclaw/openclaw | 4,124,273,460 | I_kwDOQb6kR87102s0 | 53,226 | https://github.com/openclaw/openclaw/issues/53226 | https://api.github.com/repos/openclaw/openclaw/issues/53226 | [Windows] exec launcher broken on Windows-native: all commands quoted as PS string literals, nothing executes | ### Bug type
Regression (worked before, now fails)
### Summary
Windows-native exec launcher wraps every token in single quotes and passes them to PowerShell as bare string expressions instead of invoking the command.
### Steps to reproduce
1. On Windows-native install, run `openclaw status` via exec.
2. Observe `P... | closed | completed | true | 3 | [
"bug",
"regression"
] | [] | 2026-03-23T23:24:33Z | 2026-04-30T04:46:03Z | 2026-04-25T06:16:39Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | jim777156 | 49,817,933 | MDQ6VXNlcjQ5ODE3OTMz | User | false |
openclaw/openclaw | 4,327,331,955 | I_kwDOQb6kR88AAAABAe3Ycw | 71,426 | https://github.com/openclaw/openclaw/issues/71426 | https://api.github.com/repos/openclaw/openclaw/issues/71426 | RFC: sessions.patch plugin extension hook \u2014 let plugins handle their own patch payloads | ## Problem statement
Third-party plugins that need to react to `sessions.patch` RPCs (UI inline-button events, /command-style mutations from any channel) currently have to **patch the gateway's `applySessionsPatchToStore` handler** at `src/gateway/sessions-patch.ts:89` to add new branches. This is brittle:
- Every ho... | closed | not_planned | true | 1 | [] | [] | 2026-04-25T06:18:14Z | 2026-04-29T04:36:51Z | 2026-04-26T10:54:11Z | CONTRIBUTOR | null | 20260429T060025Z | 2026-04-29T06:00:25Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,327,333,201 | I_kwDOQb6kR88AAAABAe3dUQ | 71,427 | https://github.com/openclaw/openclaw/issues/71427 | https://api.github.com/repos/openclaw/openclaw/issues/71427 | RFC: pi-embedded plugin tool-event subscription \u2014 stable seam for tool-call observation | ## Problem statement
Plugins that need to observe tool execution events (tool start/end/persist, with the agent context) currently have to patch `src/agents/pi-embedded-subscribe.handlers.tools.ts` to add a hook callout. The file is the host's tool-execution dispatcher; modifying it requires re-baselining on every hos... | closed | completed | true | 1 | [] | [] | 2026-04-25T06:18:38Z | 2026-04-28T04:46:50Z | 2026-04-25T19:33:04Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,327,334,246 | I_kwDOQb6kR88AAAABAe3hZg | 71,428 | https://github.com/openclaw/openclaw/issues/71428 | https://api.github.com/repos/openclaw/openclaw/issues/71428 | RFC: allowConversationAccess as plugin manifest field (replaces 2 schema patches) | ## Problem statement
The `allowConversationAccess` flag is set in `~/.openclaw/openclaw.json` per-plugin under `plugins.entries.<id>.hooks.allowConversationAccess: true`. The flag is gated by config schema validation; non-bundled plugins must explicitly opt in or the gateway rejects their `agent_end` typed-hook subscr... | open | null | false | 2 | [] | [] | 2026-04-25T06:19:01Z | 2026-05-02T02:13:32Z | null | CONTRIBUTOR | null | 20260502T030029Z | 2026-05-02T03:00:29Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,195,115,761 | I_kwDOQb6kR876DGLx | 59,743 | https://github.com/openclaw/openclaw/issues/59743 | https://api.github.com/repos/openclaw/openclaw/issues/59743 | [Bug]: shouldSuppressMessagingToolReplies drops legitimate final reply when message tool sent media-only to same channel | ## Summary
`shouldSuppressMessagingToolReplies()` suppresses **all** post-turn final reply payloads when the `message` tool was used to send **media-only** content (e.g., an audio file) to the **same channel** that originated the message. The legitimate final text reply is silently dropped — no error is logged, no del... | open | null | false | 2 | [] | [] | 2026-04-02T14:48:33Z | 2026-05-02T03:03:07Z | null | CONTRIBUTOR | null | 20260502T060142Z | 2026-05-02T06:01:42Z | solavrc | 145,330,217 | U_kgDOCKmQKQ | User | false |
openclaw/openclaw | 4,327,338,381 | I_kwDOQb6kR88AAAABAe3xjQ | 71,432 | https://github.com/openclaw/openclaw/issues/71432 | https://api.github.com/repos/openclaw/openclaw/issues/71432 | Feature Request: Multi-Agent Group Chat Collaboration | ## Summary
Enable multiple OpenClaw agents to participate in a shared conversation space (group chat), where they can observe each other's messages, collaborate on tasks, and produce a unified result — similar to how humans collaborate in a group chat.
## Problem
Today's `agentToAgent` mechanism is **point-to-point*... | closed | not_planned | true | 2 | [] | [] | 2026-04-25T06:20:30Z | 2026-04-30T04:38:44Z | 2026-04-26T05:27:34Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | xiaote007 | 279,194,332 | U_kgDOEKQq3A | User | false |
openclaw/openclaw | 4,075,309,045 | I_kwDOQb6kR87y6Ef1 | 45,940 | https://github.com/openclaw/openclaw/issues/45940 | https://api.github.com/repos/openclaw/openclaw/issues/45940 | False negative from `openclaw gateway probe` on Windows | ### Bug type
Regression (worked before, now fails)
### Summary
After updating to `OpenClaw 2026.3.13`, `openclaw gateway probe` reports the local gateway as unreachable, but the gateway is actually working.
### Environment
- OpenClaw: `2026.3.13 (61d171a)`
- OS: `Windows 10.0.26200 x64`
- Node: `24.14.0`
- Install ... | closed | completed | true | 8 | [
"bug",
"close:already-fixed",
"regression"
] | [] | 2026-03-14T09:35:45Z | 2026-04-30T04:45:51Z | 2026-04-25T06:22:31Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | German48 | 61,348,844 | MDQ6VXNlcjYxMzQ4ODQ0 | User | false |
openclaw/openclaw | 4,155,593,261 | I_kwDOQb6kR873sVIt | 56,010 | https://github.com/openclaw/openclaw/issues/56010 | https://api.github.com/repos/openclaw/openclaw/issues/56010 | [Bug]: Telegram voice notes are not transcribed; inbound message arrives only as <media:audio> without media context | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
### Summary
Telegram voice notes are not transcribed even with valid `tools.media.audio.models` / OpenAI Whisper configuration, because inbound Telegram voice/audio messages appear to reach the agent only as `<media:audio>` ... | closed | not_planned | true | 1 | [
"bug",
"close:duplicate",
"dedupe:child",
"regression"
] | [
"vincentkoc"
] | 2026-03-27T20:02:27Z | 2026-04-30T04:45:49Z | 2026-04-25T06:25:52Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | repy007 | 204,890,045 | U_kgDODDZfvQ | User | false |
openclaw/openclaw | 4,157,022,126 | I_kwDOQb6kR873xx-u | 56,124 | https://github.com/openclaw/openclaw/issues/56124 | https://api.github.com/repos/openclaw/openclaw/issues/56124 | Bug: Audio preflight transcription not triggering in mention-gated Telegram groups | ## Environment
- OpenClaw 2026.3.24 (stable)
- macOS 26.4 (arm64), Node 25.8.2
- Channel: Telegram group with `requireMention: true`
- Audio config: `tools.media.audio.enabled: true`, model `openai/gpt-4o-mini-transcribe`
## Expected behavior
Per docs (`nodes/audio.md` → "Mention Detection in Groups"), when `requireMe... | closed | not_planned | true | 1 | [
"close:duplicate",
"dedupe:child"
] | [
"vincentkoc"
] | 2026-03-28T02:29:25Z | 2026-04-30T04:45:48Z | 2026-04-25T06:25:58Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | mdura-spec | 262,993,836 | U_kgDOD6z3rA | User | false |
openclaw/openclaw | 4,138,405,603 | I_kwDOQb6kR872qw7j | 54,695 | https://github.com/openclaw/openclaw/issues/54695 | https://api.github.com/repos/openclaw/openclaw/issues/54695 | [Bug] STT transcription not working on Telegram (Groq) — broken since 2026.3.12, still broken on 2026.3.24 | ## Description
Audio transcription for inbound voice messages (Telegram) stopped working after updating from pre-2026.3.12. Persists through 2026.3.23-2 and 2026.3.24.
Related to #46326 (same issue on Signal).
## Environment
- **OpenClaw:** 2026.3.24 (cff6dc9)
- **OS:** Linux 6.8.0-106-generic (x64), Node v22.22.1
... | closed | not_planned | true | 2 | [
"close:duplicate",
"dedupe:child"
] | [
"vincentkoc"
] | 2026-03-25T20:44:21Z | 2026-04-30T04:45:48Z | 2026-04-25T06:26:05Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | imydal | 48,844,067 | MDQ6VXNlcjQ4ODQ0MDY3 | User | false |
openclaw/openclaw | 4,143,375,518 | I_kwDOQb6kR8729uSe | 55,145 | https://github.com/openclaw/openclaw/issues/55145 | https://api.github.com/repos/openclaw/openclaw/issues/55145 | WhatsApp: TTS tool returns success but audio not delivered when auto=inbound | ## Summary
When `messages.tts.auto` is set to `"inbound"` and a voice note is received on WhatsApp, the `tts` tool is called and returns `"Generated audio reply."` successfully, but no audio message is ever delivered to WhatsApp. The agent then replies `NO_REPLY` (suppressing duplicate text), leaving the user with no ... | closed | not_planned | true | 1 | [
"close:duplicate",
"dedupe:child"
] | [
"vincentkoc"
] | 2026-03-26T12:20:35Z | 2026-04-30T04:45:45Z | 2026-04-25T06:26:21Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Barclay1895 | 197,243,493 | U_kgDOC8GyZQ | User | false |
openclaw/openclaw | 4,327,342,024 | I_kwDOQb6kR88AAAABAe3_yA | 71,433 | https://github.com/openclaw/openclaw/issues/71433 | https://api.github.com/repos/openclaw/openclaw/issues/71433 | [Bug]: Streamable HTTP: missing required Accept header causes failures with strict MCP servers (e.g. 智谱/BigModel) | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Bug Description
mcporter fails to communicate with MCP servers that strictly enforce the Streamable HTTP transport spec. Specifically, 智谱
(BigModel/ZhipuAI) MCP services (e.g. web_search_prime, web_reader) are broken.
Root... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-25T06:22:04Z | 2026-04-30T04:45:45Z | 2026-04-25T06:32:15Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | YuhangJi | 52,953,718 | MDQ6VXNlcjUyOTUzNzE4 | User | false |
openclaw/openclaw | 4,327,371,231 | I_kwDOQb6kR88AAAABAe5x3w | 71,443 | https://github.com/openclaw/openclaw/issues/71443 | https://api.github.com/repos/openclaw/openclaw/issues/71443 | Integration: AIP Agent Guard -- skill signing, capability manifests, and audit for OpenClaw | ## Summary
[AIP Agent Guard](https://github.com/sunilp/aip-openclaw) is an open-source security plugin for OpenClaw that adds verifiable skill author identity, capability manifests with runtime enforcement, and an audit trail using the [Agent Identity Protocol](https://sunilprakash.com/aip/).
Available on ClawHub: `c... | closed | not_planned | true | 2 | [] | [] | 2026-04-25T06:36:13Z | 2026-04-28T04:42:51Z | 2026-04-25T19:33:11Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | sunilp | 633,213 | MDQ6VXNlcjYzMzIxMw== | User | false |
openclaw/openclaw | 4,198,503,959 | I_kwDOQb6kR876QBYX | 60,093 | https://github.com/openclaw/openclaw/issues/60093 | https://api.github.com/repos/openclaw/openclaw/issues/60093 | [Feature]: voice-call plugin: Support Google Gemini Live as end-to-end audio provider (STT+LLM+TTS bypass) | ## Original Request
> Create a github feature request on openclaw to support Google Live
> **English Translation:** Add support for Google Gemini Live API in the voice-call plugin, enabling end-to-end audio conversations without the separate STT→LLM→TTS pipeline.
## Agent's Two Cents (could be wrong)
> Everything b... | closed | completed | true | 2 | [] | [] | 2026-04-03T05:41:13Z | 2026-04-30T04:45:44Z | 2026-04-25T06:36:31Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | xinbenlv | 640,325 | MDQ6VXNlcjY0MDMyNQ== | User | false |
openclaw/openclaw | 3,893,572,477 | I_kwDOQb6kR87oEzN9 | 8,355 | https://github.com/openclaw/openclaw/issues/8355 | https://api.github.com/repos/openclaw/openclaw/issues/8355 | Feature: Streaming TTS pipeline for voice calls (sentence-level LLM→TTS→audio) | ## Problem
The current voice call pipeline is fully batched at every stage:
1. LLM generates the **complete** text response before anything is sent to TTS
2. TTS (ElevenLabs) synthesizes the **entire** audio buffer before playback begins
3. Only then does audio stream back to the caller via Twilio Media Streams
This... | open | null | false | 4 | [
"enhancement",
"channel: voice-call"
] | [] | 2026-02-03T22:39:54Z | 2026-05-05T00:08:21Z | null | NONE | null | 20260505T030055Z | 2026-05-05T03:00:55Z | pilkster | 5,051,408 | MDQ6VXNlcjUwNTE0MDg= | User | false |
openclaw/openclaw | 4,175,530,255 | I_kwDOQb6kR8744YkP | 58,114 | https://github.com/openclaw/openclaw/issues/58114 | https://api.github.com/repos/openclaw/openclaw/issues/58114 | Feature: Add [[tts:voiceId=...]] directive support to telephony TTS path | # Feature Request: Add `[[tts:voiceId=...]]` directive support to the telephony TTS path
## Summary
The voice-call plugin's telephony TTS path does not parse `[[tts:...]]` directives, making mid-call voice switching impossible. The directive parsing infrastructure already exists in the message channel TTS path — it j... | closed | completed | true | 3 | [
"enhancement",
"channel: voice-call"
] | [] | 2026-03-31T03:51:22Z | 2026-05-04T04:35:18Z | 2026-05-01T21:49:45Z | NONE | null | 20260504T060031Z | 2026-05-04T06:00:31Z | legonhilltech-jpg | 272,396,537 | U_kgDOEDxw-Q | User | false |
openclaw/openclaw | 4,161,590,255 | I_kwDOQb6kR874DNPv | 56,604 | https://github.com/openclaw/openclaw/issues/56604 | https://api.github.com/repos/openclaw/openclaw/issues/56604 | [Feature]: Per-number voice and persona routing for voice-call plugin | ### Summary
Add per-phone-number routing for voice calls, allowing different voices, personas, and system prompts based on which Twilio/Telnyx number was dialed.
### Problem to solve
Currently the voice-call plugin has global settings for `tts.voice`, `responseSystemPrompt`, `responseModel`, and `inboundGreeting`. ... | closed | completed | true | 3 | [
"enhancement",
"channel: voice-call"
] | [] | 2026-03-28T20:37:08Z | 2026-05-05T04:28:02Z | 2026-05-02T08:45:24Z | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | healthstatus | 1,516,140 | MDQ6VXNlcjE1MTYxNDA= | User | false |
openclaw/openclaw | 4,072,126,400 | I_kwDOQb6kR87yt7fA | 45,280 | https://github.com/openclaw/openclaw/issues/45280 | https://api.github.com/repos/openclaw/openclaw/issues/45280 | Feature request: per-call session scope for voice-call plugin | ## Problem
The voice-call plugin keys sessions by caller phone number (`voice:${normalizedPhone}` in `response-generator.ts`). This means all calls from the same number share conversation history, causing the AI to "remember" previous calls — referencing prior appointments, caller details, etc.
For use cases like a p... | closed | completed | true | 3 | [
"enhancement",
"channel: voice-call"
] | [] | 2026-03-13T16:26:24Z | 2026-05-05T04:28:14Z | 2026-05-02T07:43:42Z | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | pondcountry | 267,987,093 | U_kgDOD_kolQ | User | false |
openclaw/openclaw | 4,189,768,189 | I_kwDOQb6kR875usn9 | 59,245 | https://github.com/openclaw/openclaw/issues/59245 | https://api.github.com/repos/openclaw/openclaw/issues/59245 | [Feature]: Outbound Task Calls — Agent makes phone calls on user's behalf (reservations, inquiries, flight changes) | ## Original Request
> 我经常需要请我的助理去打电话，比如：
> 1. 预约餐厅
> 2. 预约医院的医生
> 3. 询问一些具体问题（例如打电话给航空公司要求更换或切换航班等）
>
> 我希望这个 Agent 可以打通电话，能够识别通话中的等待状态，并获取相关信息。它可以把我授权的信息提供给对方，同时也具备灵活变通的能力。
>
> 在遇到需要主人明确指示的情况时，它会主动联系我说："Hey, I'm on the call. 我现在正在跟谁通话，需要你及时的回应。"
>
> 如果需要等待主人的回复，它能让对方先等一下。如果等待超过两三分钟我还没有回应，它会跟对方说："我主人还没来得及回应我，我稍后再给你回过去... | open | null | false | 3 | [
"enhancement",
"channel: voice-call"
] | [] | 2026-04-01T20:44:57Z | 2026-05-01T06:04:59Z | null | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | xinbenlv | 640,325 | MDQ6VXNlcjY0MDMyNQ== | User | false |
openclaw/openclaw | 3,920,881,659 | I_kwDOQb6kR87ps-f7 | 13,337 | https://github.com/openclaw/openclaw/issues/13337 | https://api.github.com/repos/openclaw/openclaw/issues/13337 | [Feature] Voice Call Plugin: Add Vapi provider | ## 🎯 作業宣言
Voice Call Plugin に **Vapi プロバイダー**を追加します。
## 📋 タスク内容
### 概要
OpenClaw Voice Call Plugin (`extensions/voice-call/`) に Vapi を新規プロバイダーとして実装します。
### 実装範囲
1. **VapiProvider クラス** (`src/providers/vapi.ts`)
- `VoiceCallProvider` インターフェース実装
- 通話開始・終了・TTS 再生メソッド
- Vapi REST API 呼び出し
2. **設定スキーマ** (`sr... | open | null | false | 5 | [
"enhancement",
"channel: voice-call"
] | [] | 2026-02-10T11:15:18Z | 2026-05-05T01:19:11Z | null | NONE | null | 20260505T030055Z | 2026-05-05T03:00:55Z | ShunsukeHayashi | 11,504,206 | MDQ6VXNlcjExNTA0MjA2 | User | false |
openclaw/openclaw | 3,918,448,396 | I_kwDOQb6kR87pjscM | 12,911 | https://github.com/openclaw/openclaw/issues/12911 | https://api.github.com/repos/openclaw/openclaw/issues/12911 | Feature: Live Voice Mode in /chat UI (Grok Voice Loop) | ## Summary
Replace the existing **Mic** button in the webchat UI (`/chat`) with a **Live** toggle that enables a real-time voice conversation loop between the user and the agent, powered by xAI's streaming voice API (Grok voice).
## Current Behavior
- The webchat has a Mic button (presumably for push-to-talk or spee... | open | null | false | 3 | [
"enhancement"
] | [] | 2026-02-09T22:25:47Z | 2026-05-05T01:19:01Z | null | NONE | null | 20260505T030055Z | 2026-05-05T03:00:55Z | haverAI | 131,009,773 | U_kgDOB88M7Q | User | false |
openclaw/openclaw | 3,899,816,675 | I_kwDOQb6kR87ocnrj | 9,401 | https://github.com/openclaw/openclaw/issues/9401 | https://api.github.com/repos/openclaw/openclaw/issues/9401 | feat(voice-web): Client-side VAD to reduce unnecessary API calls | ## Summary
Currently, `voice-web` sends all audio frames to the Gateway, which performs VAD server-side. This is inefficient:
- Audio frames sent even during silence
- Unnecessary network bandwidth and server load
- API costs incurred even when no speech is detected
## Proposed Solution
Add client-side VAD (Voice A... | closed | not_planned | true | 2 | [
"enhancement"
] | [] | 2026-02-05T05:49:57Z | 2026-04-30T04:45:40Z | 2026-04-25T06:42:44Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | code-yeongyu | 11,153,873 | MDQ6VXNlcjExMTUzODcz | User | false |
openclaw/openclaw | 4,215,488,349 | I_kwDOQb6kR877Qz9d | 62,302 | https://github.com/openclaw/openclaw/issues/62302 | https://api.github.com/repos/openclaw/openclaw/issues/62302 | [Bug]: RangeError: Maximum call stack size exceeded on startup due to talk-voice invoking JITI AST compilation on monolithic io core | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
When attempting to start the openclaw dashboard, the startup process fatally crashes (or hangs) on Windows with a RangeError: Maximum call stack size exceeded exception. The stack overflow occurs in the plugin loader during... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-07T04:50:20Z | 2026-04-30T04:45:31Z | 2026-04-25T06:45:05Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | SecretGuestZ | 24,306,845 | MDQ6VXNlcjI0MzA2ODQ1 | User | false |
openclaw/openclaw | 4,188,930,630 | I_kwDOQb6kR875rgJG | 59,192 | https://github.com/openclaw/openclaw/issues/59192 | https://api.github.com/repos/openclaw/openclaw/issues/59192 | Feature Request: Native local Whisper integration for audio transcription | ## Feature Request
### Summary
Add native support for local OpenAI Whisper audio transcription, similar to the existing Groq integration, but without requiring external API calls.
### Current Situation
- Groq audio transcription works but requires paid API access
- Local Whisper CLI works via tool but requires appro... | closed | completed | true | 1 | [] | [] | 2026-04-01T18:14:49Z | 2026-04-30T04:45:21Z | 2026-04-25T06:46:13Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | wauwi-himbeersternchen | 269,577,791 | U_kgDOEBFuPw | User | false |
openclaw/openclaw | 4,141,873,873 | I_kwDOQb6kR8723_rR | 55,052 | https://github.com/openclaw/openclaw/issues/55052 | https://api.github.com/repos/openclaw/openclaw/issues/55052 | [Bug] Telegram voice messages: media understanding audio transcription pipeline never triggered | ## Summary
Telegram voice messages are received and downloaded successfully, but the `applyMediaUnderstanding` audio transcription pipeline is never invoked. The agent receives `<media:audio>` with the raw `.ogg` file attached but no automatic transcription occurs.
**This affects all channels** (confirmed on both Tel... | closed | completed | true | 1 | [] | [] | 2026-03-26T08:47:05Z | 2026-04-30T04:45:20Z | 2026-04-25T06:46:15Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | stanleeyY | 4,137,404 | MDQ6VXNlcjQxMzc0MDQ= | User | false |
openclaw/openclaw | 4,327,348,107 | I_kwDOQb6kR88AAAABAe4Xiw | 71,435 | https://github.com/openclaw/openclaw/issues/71435 | https://api.github.com/repos/openclaw/openclaw/issues/71435 | [Bug] DeepSeek V4 thinking mode fails with 400: reasoning_content must be passed back on multi-turn tool calls | ## Bug Report
### Description
When using DeepSeek V4 models (`deepseek-v4-flash` / `deepseek-v4-pro`) with thinking mode enabled, multi-turn conversations that involve tool calls fail with a 400 error:
```
The `content[].thinking` in the thinking mode must be passed back to the API.
```
This happens on every follow... | closed | completed | true | 1 | [] | [] | 2026-04-25T06:24:43Z | 2026-04-30T04:45:15Z | 2026-04-25T06:46:40Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | leoge007 | 116,452,300 | U_kgDOBvDrzA | User | false |
openclaw/openclaw | 4,206,751,284 | I_kwDOQb6kR876ve40 | 61,217 | https://github.com/openclaw/openclaw/issues/61217 | https://api.github.com/repos/openclaw/openclaw/issues/61217 | [Feature]: Feature Request: Native Telegram Voice Message Auto-Transcription | ### Summary
Currently, OpenClaw receives Telegram voice messages as `.ogg` attachments but does not automatically transcribe them into text. Users have to manually trigger a transcription script or ask the agent to process the file. This feature request proposes adding a native `speechToText` configuration option for ... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-04-05T06:41:43Z | 2026-04-30T04:45:12Z | 2026-04-25T06:49:31Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | smc6184 | 261,246,199 | U_kgDOD5JM9w | User | false |
openclaw/openclaw | 4,192,224,575 | I_kwDOQb6kR8754EU_ | 59,510 | https://github.com/openclaw/openclaw/issues/59510 | https://api.github.com/repos/openclaw/openclaw/issues/59510 | [Feature]: Simplify exec approval process | ### Summary
The current exec command approval process in OpenClaw v2026.4.1 is overly complicated, severely impacting usability and efficiency.
### Problem to solve
1. **Tedious Per-Command Approvals**
- Need to set `/approve xxx allow-always` for each individual command
- Dozens of common commands require rep... | closed | completed | true | 16 | [
"enhancement"
] | [] | 2026-04-02T06:58:02Z | 2026-04-29T04:39:53Z | 2026-04-26T09:00:49Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | alixzao | 264,385,920 | U_kgDOD8I1gA | User | false |
openclaw/openclaw | 4,220,151,822 | I_kwDOQb6kR877imgO | 62,680 | https://github.com/openclaw/openclaw/issues/62680 | https://api.github.com/repos/openclaw/openclaw/issues/62680 | [Bug]: Telegram DM voice notes not auto-transcribed when wrapper text makes hasUserText=true | ## Summary
Related to #17101.
In Telegram DMs, voice notes can arrive in the session wrapped with queue/envelope text and a `<media:audio>` placeholder. That wrapper text appears to make `hasUserText = true`, which then prevents `transcribeFirstAudio(...)` from running, even though the semantic user payload is audio-... | closed | completed | true | 1 | [] | [] | 2026-04-07T19:37:08Z | 2026-04-30T04:45:10Z | 2026-04-25T06:51:04Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | DerFlash | 185,013 | MDQ6VXNlcjE4NTAxMw== | User | false |
openclaw/openclaw | 4,138,415,054 | I_kwDOQb6kR872qzPO | 54,696 | https://github.com/openclaw/openclaw/issues/54696 | https://api.github.com/repos/openclaw/openclaw/issues/54696 | [Bug] TTS voice replies arrive as MP3 file attachment instead of native Telegram voice message (2026.3.24) | ## Description
TTS audio replies on Telegram are delivered as MP3 file attachments (document) instead of native voice messages (voice bubble with waveform).
The file is named `voice-*.mp3` but is sent via `sendDocument`/`sendAudio` instead of `sendVoice`.
## Environment
- **OpenClaw:** 2026.3.24 (cff6dc9)
- **OS:**... | closed | completed | true | 2 | [] | [] | 2026-03-25T20:46:06Z | 2026-04-30T04:45:08Z | 2026-04-25T06:51:40Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | imydal | 48,844,067 | MDQ6VXNlcjQ4ODQ0MDY3 | User | false |
openclaw/openclaw | 4,196,368,022 | I_kwDOQb6kR876H36W | 59,864 | https://github.com/openclaw/openclaw/issues/59864 | https://api.github.com/repos/openclaw/openclaw/issues/59864 | [Bug]: Telegram voice note transcription fails with Deepgram — audio arrives as raw <media:audio> without transcript | ### Summary
Telegram voice notes are no longer transcribed via the configured Deepgram provider. Deepgram transcription appears to be **completely broken** — it never runs.
When transcription does appear to work, it is because OpenClaw silently falls back to the **local Whisper CLI** (via bash). This fallback only w... | closed | completed | true | 1 | [] | [] | 2026-04-02T18:50:16Z | 2026-04-30T04:45:05Z | 2026-04-25T06:52:47Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | coccoinomane | 3,920,436 | MDQ6VXNlcjM5MjA0MzY= | User | false |
openclaw/openclaw | 4,327,357,291 | I_kwDOQb6kR88AAAABAe47aw | 71,438 | https://github.com/openclaw/openclaw/issues/71438 | https://api.github.com/repos/openclaw/openclaw/issues/71438 | [Feishu] group_topic session splits when different root_id formats (omt_ vs om_) used for same topic | ## Bug: Feishu group_topic session splits into two sessions for same topic
**Environment**
- OpenClaw (latest)
- Channel: Feishu/Lark
- Config: `groupSessionScope: "group_topic"` on a specific group
**Steps to Reproduce**
1. Configure a Feishu group with `groupSessionScope: "group_topic"` and `replyInThread: "enable... | closed | completed | true | 1 | [] | [] | 2026-04-25T06:28:41Z | 2026-04-30T04:45:05Z | 2026-04-25T06:53:43Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | 1335848090 | 74,958,878 | MDQ6VXNlcjc0OTU4ODc4 | User | false |
openclaw/openclaw | 4,233,660,278 | I_kwDOQb6kR878WId2 | 63,851 | https://github.com/openclaw/openclaw/issues/63851 | https://api.github.com/repos/openclaw/openclaw/issues/63851 | Groq audio transcription sends JSON instead of multipart/form-data (v2026.4.9) | ## Bug Description
Groq audio transcription via `tools.media.audio.models` with `{"provider": "groq"}` fails because OpenClaw sends the request as JSON instead of `multipart/form-data`. Groq's `/audio/transcriptions` endpoint returns:
```
Error: request Content-Type isn't multipart/form-data
Type: invalid_request_err... | closed | completed | true | 1 | [] | [] | 2026-04-09T16:40:09Z | 2026-04-30T04:45:04Z | 2026-04-25T06:53:54Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | darker0n | 11,329,474 | MDQ6VXNlcjExMzI5NDc0 | User | false |
openclaw/openclaw | 4,219,275,103 | I_kwDOQb6kR877fQdf | 62,605 | https://github.com/openclaw/openclaw/issues/62605 | https://api.github.com/repos/openclaw/openclaw/issues/62605 | Groq audio transcription fails with invalid_request_error: request Content-Type isn't multipart/form-data | Hi, I’m seeing what looks like a Groq audio transcription integration issue in OpenClaw.
## Summary
Telegram voice transcription works in my setup with Deepgram, but Groq appears to fail on the audio transcription request itself.
Groq returns:
`Error: request Content-Type isn't multipart/form-data`
`Type: invalid_re... | closed | completed | true | 2 | [] | [] | 2026-04-07T16:53:59Z | 2026-04-30T04:45:03Z | 2026-04-25T06:53:56Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | arassypnov | 5,262,022 | MDQ6VXNlcjUyNjIwMjI= | User | false |
openclaw/openclaw | 4,214,781,597 | I_kwDOQb6kR877OHad | 62,209 | https://github.com/openclaw/openclaw/issues/62209 | https://api.github.com/repos/openclaw/openclaw/issues/62209 | [Bug]: Telegram voice messages not transcribed via Groq Whisper after multi-agent config (Docker) | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Telegram voice messages (.ogg) are received and saved to media/inbound/ correctly, but Groq Whisper transcription is never triggered. Agent responds asking user to send text instead. No audio-related logs appear even with OP... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-07T00:41:18Z | 2026-04-30T04:45:01Z | 2026-04-25T06:54:00Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | danielsgar | 35,976,843 | MDQ6VXNlcjM1OTc2ODQz | User | false |
openclaw/openclaw | 4,119,932,068 | I_kwDOQb6kR871kSyk | 52,787 | https://github.com/openclaw/openclaw/issues/52787 | https://api.github.com/repos/openclaw/openclaw/issues/52787 | Bug: Image upload in Control UI webchat not working | ## Description
When uploading images via the Control UI (webchat) attachment button, the images disappear and are not received by the agent.
## Steps to Reproduce
1. Open Control UI at http://127.0.0.1:18789/
2. Click the attachment button (📎) in the chat input area
3. Select an image file
4. The image disappears f... | closed | completed | true | 7 | [
"close:duplicate",
"close:already-fixed",
"dedupe:child"
] | [
"vincentkoc"
] | 2026-03-23T10:45:56Z | 2026-04-30T04:45:00Z | 2026-04-25T06:09:00Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | moliclaw1212 | 270,327,306 | U_kgDOEBzeCg | User | false |
openclaw/openclaw | 4,233,163,798 | I_kwDOQb6kR878UPQW | 63,791 | https://github.com/openclaw/openclaw/issues/63791 | https://api.github.com/repos/openclaw/openclaw/issues/63791 | [Bug] Groq STT multipart submission breaks in runtime/undici fetch path; native fetch succeeds | ## Summary
Groq audio transcription requests in OpenClaw can fail even when the request is correctly constructed as `FormData`.
In my testing, the request reaches OpenClaw's runtime fetch layer with:
- correct Groq transcription URL
- correct auth header
- `body instanceof FormData === true`
But when sent through t... | closed | completed | true | 1 | [] | [] | 2026-04-09T15:04:07Z | 2026-04-30T04:44:59Z | 2026-04-25T06:58:28Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Theo-jobs | 177,595,285 | U_kgDOCpXjlQ | User | false |
openclaw/openclaw | 4,225,448,023 | I_kwDOQb6kR8772zhX | 63,201 | https://github.com/openclaw/openclaw/issues/63201 | https://api.github.com/repos/openclaw/openclaw/issues/63201 | Audio transcription silently fails with HTTP 400 when using undici's fetch dispatcher (FormData Content-Type not set) | ## Summary
After fixing the SSRF block (#63132 / #63147), audio transcription still silently fails for self-hosted STT providers. The server receives a malformed multipart body and returns HTTP 400 \`{"error":"No file part in the request"}\`. No error is visible to the user — the transcript comes back as \`undefined\`... | closed | completed | true | 3 | [] | [] | 2026-04-08T14:53:34Z | 2026-04-30T04:44:57Z | 2026-04-25T06:58:30Z | CONTRIBUTOR | null | 20260430T060018Z | 2026-04-30T06:00:18Z | jhsmith409 | 198,252,919 | U_kgDOC9EZdw | User | false |
openclaw/openclaw | 4,226,748,278 | I_kwDOQb6kR8777w92 | 63,299 | https://github.com/openclaw/openclaw/issues/63299 | https://api.github.com/repos/openclaw/openclaw/issues/63299 | [Bug]: OpenClaw STT multipart FormData fails with 422 against OpenAI-compatible /audio/transcriptions endpoint, while manual multipart works | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
OpenClaw STT breaks multipart.
When using an OpenAI-compatible /v1/audio/transcriptions endpoint (self-hosted FastWhisper), OpenClaw consistently returns:
HTTP 422
failed parsing request body
At the same time:
• same au... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-08T18:30:03Z | 2026-04-30T04:44:56Z | 2026-04-25T06:58:34Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | utya | 11,736,439 | MDQ6VXNlcjExNzM2NDM5 | User | false |
openclaw/openclaw | 4,085,311,223 | I_kwDOQb6kR87zgOb3 | 48,554 | https://github.com/openclaw/openclaw/issues/48554 | https://api.github.com/repos/openclaw/openclaw/issues/48554 | fix: audio transcription sends [object FormData] when HTTP proxy is configured (Node 22+ FormData/undici mismatch) | ## Bug Type
Behavior bug
## Summary
When `HTTPS_PROXY` (or `HTTP_PROXY`) is set in the environment, audio transcription requests (and batch file uploads) silently fail. The request body is sent as the literal string `[object FormData]` with `Content-Type: text/plain;charset=UTF-8` instead of proper `multipart/form-d... | closed | completed | true | 5 | [] | [] | 2026-03-16T23:23:37Z | 2026-05-05T04:28:07Z | 2026-05-02T08:22:13Z | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | dco5 | 3,612,771 | MDQ6VXNlcjM2MTI3NzE= | User | false |
openclaw/openclaw | 4,208,990,991 | I_kwDOQb6kR8764BsP | 61,590 | https://github.com/openclaw/openclaw/issues/61590 | https://api.github.com/repos/openclaw/openclaw/issues/61590 | Telegram voice works in explicit tests but is inconsistent in normal reply flow | ### Summary
On this OpenClaw instance, Telegram voice output appears technically available but operationally inconsistent in normal assistant turns.
### Environment
- OpenClaw: 2026.4.2
- Telegram channel enabled and functioning for normal text delivery
- TTS/voice route has been observed working at least in explicit ... | open | null | false | 2 | [] | [] | 2026-04-06T01:10:35Z | 2026-05-02T07:02:02Z | null | NONE | null | 20260502T090030Z | 2026-05-02T09:00:30Z | tadeogutierrez | 16,327,260 | MDQ6VXNlcjE2MzI3MjYw | User | false |
openclaw/openclaw | 4,220,497,465 | I_kwDOQb6kR877j645 | 62,698 | https://github.com/openclaw/openclaw/issues/62698 | https://api.github.com/repos/openclaw/openclaw/issues/62698 | Feature: Send audio/media files from agent to Telegram channel | ## Problem
When an agent processes audio files (e.g., extracts audio from YouTube videos, generates audio, or processes voice notes), there is no way to send the audio file back to the user through Telegram.
## Current behavior
- Agents can receive audio files from users (`<media:audio>`)
- Agents can send images vi... | closed | completed | true | 1 | [] | [] | 2026-04-07T20:44:28Z | 2026-04-30T04:44:54Z | 2026-04-25T07:00:17Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Javisandomcoder | 28,111,502 | MDQ6VXNlcjI4MTExNTAy | User | false |
openclaw/openclaw | 4,115,170,457 | I_kwDOQb6kR871SISZ | 52,125 | https://github.com/openclaw/openclaw/issues/52125 | https://api.github.com/repos/openclaw/openclaw/issues/52125 | TTS tags not processed in cron announce delivery | ## Summary
`[[tts]]` and `[[tts:text]]` tags in agent replies are not processed when delivered via cron job `announce` mode. The tags render as raw text in Telegram instead of generating voice notes.
## Steps to Reproduce
1. Configure TTS in `openclaw.json`:
```json
{
"messages": {
"tts": {
"auto": "tagg... | closed | completed | true | 5 | [] | [] | 2026-03-22T07:02:47Z | 2026-05-04T04:34:37Z | 2026-05-02T01:56:29Z | NONE | null | 20260504T060031Z | 2026-05-04T06:00:31Z | kenchen3000 | 9,312,409 | MDQ6VXNlcjkzMTI0MDk= | User | false |
openclaw/openclaw | 4,327,334,674 | I_kwDOQb6kR88AAAABAe3jEg | 71,430 | https://github.com/openclaw/openclaw/issues/71430 | https://api.github.com/repos/openclaw/openclaw/issues/71430 | [Bug] Embedded agent context overflows 200k limit without auto-compaction firing | ## Summary
A long-running embedded agent (Anthropic Claude Opus 4.7, 200k context window) reached and exceeded its hard context limit (~300k observed total tokens, 150% of cap) without the documented auto-compaction routine triggering. Result: model responses get silently truncated, completion events do not fire, downs... | closed | not_planned | true | 1 | [] | [] | 2026-04-25T06:19:10Z | 2026-04-30T04:44:52Z | 2026-04-25T07:00:30Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | sync-tachikoma | 278,301,682 | U_kgDOEJaL8g | User | false |
openclaw/openclaw | 4,160,255,475 | I_kwDOQb6kR873-HXz | 56,444 | https://github.com/openclaw/openclaw/issues/56444 | https://api.github.com/repos/openclaw/openclaw/issues/56444 | message tool: WhatsApp media sends return 'unsupported channel' error | # WhatsApp Media Sending Bug Report
## Issue Description
The `message` tool incorrectly returns `"unsupported channel: whatsapp"` when attempting to send media (audio/image/video) via WhatsApp, despite the channel being fully configured and capable of media transmission.
## Environment
- OpenClaw version: 2026.3.24
-... | closed | completed | true | 3 | [] | [] | 2026-03-28T14:48:44Z | 2026-04-30T04:44:49Z | 2026-04-25T07:03:39Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | AndlerRL | 35,474,730 | MDQ6VXNlcjM1NDc0NzMw | User | false |
openclaw/openclaw | 4,197,099,510 | I_kwDOQb6kR876Kqf2 | 59,919 | https://github.com/openclaw/openclaw/issues/59919 | https://api.github.com/repos/openclaw/openclaw/issues/59919 | Bug: TTS tool reports success but no audio file is generated or delivered on WhatsApp | ## Description
The `tts` tool returns `"Generated audio reply."` but no audio file is actually produced or delivered to the WhatsApp chat. The user receives nothing.
## Environment
- **OpenClaw version:** 2026.4.2 (also reproduced on 2026.4.1)
- **Channel:** WhatsApp (Baileys)
- **OS:** Ubuntu Linux 6.17.0-1007-aws ... | closed | completed | true | 1 | [] | [] | 2026-04-02T21:34:54Z | 2026-04-30T04:44:49Z | 2026-04-25T07:03:48Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | fray-ai | 261,794,680 | U_kgDOD5qreA | User | false |
openclaw/openclaw | 4,214,036,887 | I_kwDOQb6kR877LRmX | 62,122 | https://github.com/openclaw/openclaw/issues/62122 | https://api.github.com/repos/openclaw/openclaw/issues/62122 | TTS tool audio dropped when model replies with NO_REPLY + per-channel-peer routing fails for WhatsApp | ## Summary
When the TTS tool generates audio and the model replies with `NO_REPLY` (as instructed by the TTS tool description), the outbound reply filter strips the entire payload including the queued TTS media. Additionally, `per-channel-peer` DM sessions route through the wrong dispatch pipeline, silently dropping W... | closed | completed | true | 3 | [] | [] | 2026-04-06T20:58:28Z | 2026-04-28T04:56:09Z | 2026-04-25T17:35:43Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | alexbixor | 29,897,901 | MDQ6VXNlcjI5ODk3OTAx | User | false |
openclaw/openclaw | 4,191,166,084 | I_kwDOQb6kR8750B6E | 59,372 | https://github.com/openclaw/openclaw/issues/59372 | https://api.github.com/repos/openclaw/openclaw/issues/59372 | [Feature]: Graceful config reload — don't abort active agent sessions | ## Problem
When an agent dynamically creates a new Discord channel (e.g., by updating the OpenClaw config to add a channel entry), the config change triggers a hot-reload (SIGUSR1 in hybrid mode). This **aborts the agent's own active session**, breaking any multi-step workflow that was in progress.
### Real-world sce... | closed | completed | true | 2 | [] | [] | 2026-04-02T03:17:36Z | 2026-04-28T04:58:51Z | 2026-04-25T17:05:30Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | kagura-agent | 268,167,063 | U_kgDOD_vnlw | User | false |
openclaw/openclaw | 4,133,155,332 | I_kwDOQb6kR872WvIE | 54,321 | https://github.com/openclaw/openclaw/issues/54321 | https://api.github.com/repos/openclaw/openclaw/issues/54321 | feat(gateway): graceful config/cron reload without restart | ## Problem
When agents need to modify cron jobs or gateway config at runtime, the only option is to stop and restart the gateway. This kills all active sessions, drops channel connections, and makes the agent unresponsive until the gateway comes back up.
Common scenarios that currently require a restart:
- Editing cr... | closed | completed | true | 2 | [] | [] | 2026-03-25T07:09:01Z | 2026-04-28T05:08:02Z | 2026-04-25T14:14:50Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | agent1-tao | 269,897,938 | U_kgDOEBZQ0g | User | false |
openclaw/openclaw | 4,074,672,706 | I_kwDOQb6kR87y3pJC | 45,746 | https://github.com/openclaw/openclaw/issues/45746 | https://api.github.com/repos/openclaw/openclaw/issues/45746 | Resource: Comprehensive Chinese Tutorial for OpenClaw Users (42.3万字完整教程) | # Resource: Comprehensive Chinese Tutorial for OpenClaw Users (42.3万字完整教程)
## 📚 Tutorial Overview
I have created a comprehensive Chinese tutorial for OpenClaw users, which has been actively maintained and updated to cover the latest OpenClaw v2026.3.12 features.
### Tutorial Highlights
- **Repository**: https://gi... | closed | completed | true | 2 | [] | [] | 2026-03-14T04:47:05Z | 2026-04-30T04:44:47Z | 2026-04-25T07:10:57Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | xianyu110 | 53,889,513 | MDQ6VXNlcjUzODg5NTEz | User | false |
openclaw/openclaw | 4,327,451,572 | I_kwDOQb6kR88AAAABAe-rtA | 71,452 | https://github.com/openclaw/openclaw/issues/71452 | https://api.github.com/repos/openclaw/openclaw/issues/71452 | feat(message): list chat / list messages should support pagination instead of hardcoded 25 limit | ## Problem
`renderMessageList` in `message-format` truncates results with `.slice(0, 25)`, so `message(action=read)` and `message(action=list-pins)` can never return more than 25 items regardless of what the user requests.
Similarly, the `channel-list` action for MS Teams chats may also be limited.
## Expected Behav... | open | null | false | 3 | [] | [] | 2026-04-25T07:12:24Z | 2026-05-02T02:13:34Z | null | NONE | null | 20260502T030029Z | 2026-05-02T03:00:29Z | JerryTao-AI | 221,452,556 | U_kgDODTMZDA | User | false |
openclaw/openclaw | 4,327,334,638 | I_kwDOQb6kR88AAAABAe3i7g | 71,429 | https://github.com/openclaw/openclaw/issues/71429 | https://api.github.com/repos/openclaw/openclaw/issues/71429 | [Bug] Telegram gateway drops in-flight messages on sendChatAction network failure during hot reload | ## Summary
During gateway hot reload, `sendChatAction` calls fail with network errors and any in-flight outbound message that has not yet been delivered to Telegram is silently dropped. The agent appears to have responded (text exists in internal INFO stream and is visible in webchat UI), but the user on Telegram never... | open | null | false | 3 | [] | [] | 2026-04-25T06:19:09Z | 2026-05-02T03:03:55Z | null | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | sync-tachikoma | 278,301,682 | U_kgDOEJaL8g | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.