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 | 4,034,909,221 | I_kwDOQb6kR87wf9Ql | 38,069 | https://github.com/openclaw/openclaw/issues/38069 | https://api.github.com/repos/openclaw/openclaw/issues/38069 | [Feature]: cron job execution retry with configurable delay after transient failure | ### Summary
Add configurable execution retry for cron jobs that re-runs failed executions after a delay, rather than waiting for the next scheduled time.
### Problem to solve
When a recurring cron job fails due to a transient error (network outage, provider timeout, OAuth refresh failure during connectivity loss), t... | closed | not_planned | true | 2 | [
"enhancement"
] | [] | 2026-03-06T15:53:20Z | 2026-05-02T04:27:15Z | 2026-04-29T20:26:23Z | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | rkcollins | 104,794,829 | U_kgDOBj8KzQ | User | false |
openclaw/openclaw | 4,035,266,562 | I_kwDOQb6kR87whUgC | 38,204 | https://github.com/openclaw/openclaw/issues/38204 | https://api.github.com/repos/openclaw/openclaw/issues/38204 | Support for the new Grok 4.2 Multi Model | ### Summary
# OpenClaw + xAI grok-4.20: HTTP 400 on Chat Completions
## Problem
Adding `grok-4.20-multi-agent-experimental-beta-0304` to your OpenClaw config and using it via `openclaw tui` results in an **HTTP 400 (no body)** error.
## Root Cause
The `grok-4.20-multi-agent-experimental-beta-0304` model **exists**... | closed | not_planned | false | 6 | [
"enhancement"
] | [] | 2026-03-06T17:14:28Z | 2026-05-03T21:24:15Z | 2026-05-03T21:24:15Z | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | gocyclic249 | 6,832,918 | MDQ6VXNlcjY4MzI5MTg= | User | false |
openclaw/openclaw | 4,036,279,849 | I_kwDOQb6kR87wlL4p | 38,346 | https://github.com/openclaw/openclaw/issues/38346 | https://api.github.com/repos/openclaw/openclaw/issues/38346 | [Feature]: Telegram listen to all messages but triggered by few | ### Summary
A few that allows telegram bot to listen to all conversation in group chat so as to get users ID noted even without their name on the allow list and only the allow list can trigger the bot
### Problem to solve
This feature if implemented makes the bot recognized the users in the group chat and Identify t... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-03-06T21:37:14Z | 2026-05-04T10:02:44Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | 0xCheetah1 | 238,049,946 | U_kgDODjBamg | User | false |
openclaw/openclaw | 4,036,556,370 | I_kwDOQb6kR87wmPZS | 38,382 | https://github.com/openclaw/openclaw/issues/38382 | https://api.github.com/repos/openclaw/openclaw/issues/38382 | [Feature]: Recursively scan symlinked directories in workspace for agent context injection | ### Summary
Currently, workspace files (e.g., SOUL.md, USER.md) are injected into agent context automatically, but symlinked directories don't seem to be scanned recursively. This limits sharing external content (like an Obsidian note vault) without duplicating files, which is inefficient for large corpora.
### Prob... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-06T23:04:11Z | 2026-04-28T04:43:34Z | 2026-04-26T01:34:58Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | LorinRicker | 1,362,449 | MDQ6VXNlcjEzNjI0NDk= | User | false |
openclaw/openclaw | 4,329,757,343 | I_kwDOQb6kR88AAAABAhLanw | 71,876 | https://github.com/openclaw/openclaw/issues/71876 | https://api.github.com/repos/openclaw/openclaw/issues/71876 | Expand Windows node default allowlist for safe declared companion commands | Windows nodes are currently treated like Linux/headless exec hosts in `src/gateway/node-command-policy.ts`:
```ts
windows: [...SYSTEM_COMMANDS]
```
That means the gateway filters out safe companion-app commands that a Windows node explicitly declares, including `canvas.*`, `camera.list`, `location.get`, and `screen.s... | closed | completed | true | 0 | [
"maintainer"
] | [] | 2026-04-26T01:37:36Z | 2026-05-01T04:44:16Z | 2026-04-28T09:01:22Z | CONTRIBUTOR | null | 20260501T113057Z | 2026-05-01T11:30:57Z | shanselman | 2,892 | MDQ6VXNlcjI4OTI= | User | true |
openclaw/openclaw | 4,329,745,765 | I_kwDOQb6kR88AAAABAhKtZQ | 71,875 | https://github.com/openclaw/openclaw/issues/71875 | https://api.github.com/repos/openclaw/openclaw/issues/71875 | OpenRouter calls stamped as "pi" instead of "OpenClaw" — pi-coding-agent SDK overrides attribution module | ## Summary
OpenClaw ships a provider-attribution module that explicitly stamps
`X-OpenRouter-Title: OpenClaw` on outbound OpenRouter requests and even
flags those headers as protected from caller override. But on the wire,
every OpenRouter request from `openclaw-gateway` is stamped as **`pi`**
because the bundled `@ma... | closed | completed | true | 2 | [] | [] | 2026-04-26T01:30:16Z | 2026-04-28T04:43:33Z | 2026-04-26T01:34:28Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | protostatis | 121,765,396 | U_kgDOB0H-FA | User | false |
openclaw/openclaw | 4,036,928,176 | I_kwDOQb6kR87wnqKw | 38,458 | https://github.com/openclaw/openclaw/issues/38458 | https://api.github.com/repos/openclaw/openclaw/issues/38458 | [Feature]: Option to suppress tool error output from reaching chat surfaces | ### Summary
Add a per-channel config option (e.g. `streamToolErrors: false`) to prevent raw tool error JSON from being streamed to chat surfaces (Telegram, Discord, etc.) during agent responses.
### Problem to solve
When a tool call fails during an agent response (e.g., `message send` to an invalid target, `exec` pe... | closed | not_planned | false | 1 | [
"enhancement"
] | [] | 2026-03-07T01:33:08Z | 2026-05-03T21:49:36Z | 2026-05-03T21:49:36Z | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | winjohn777-glitch | 229,473,402 | U_kgDODa18eg | User | false |
openclaw/openclaw | 4,037,344,142 | I_kwDOQb6kR87wpPuO | 38,573 | https://github.com/openclaw/openclaw/issues/38573 | https://api.github.com/repos/openclaw/openclaw/issues/38573 | [Feature]: Allow topic/forum cron deliveries to use main-session announce/agent-turn flow instead of forced direct delivery | ### Summary
Add a config or per-cron option to let topic/forum targets use the same announce → main-agent-turn behavior that currently works in DMs, instead of always forcing direct delivery.
### Problem to solve
Right now isolated cron jobs targeting topic/forum chats are forced into direct delivery when the resolv... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-03-07T04:40:44Z | 2026-05-04T06:15:29Z | null | NONE | null | 20260504T090105Z | 2026-05-04T09:01:05Z | wonrax | 28,887,874 | MDQ6VXNlcjI4ODg3ODc0 | User | false |
openclaw/openclaw | 4,037,671,736 | I_kwDOQb6kR87wqfs4 | 38,654 | https://github.com/openclaw/openclaw/issues/38654 | https://api.github.com/repos/openclaw/openclaw/issues/38654 | [Feature]: Support multiple OpenClaw instances in same Telegram group | ### Summary
Enable multiple OpenClaw instances to collaborate in the same Telegram group
### Problem to solve
In a Telegram group with multiple OpenClaw instances (different bots), bots cannot see each other's messages. This prevents:
1. One bot @mentioning another bot in the group
2. Collaborative work between mul... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-03-07T06:59:12Z | 2026-04-28T04:43:21Z | 2026-04-26T01:46:03Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | adminlove520 | 32,920,883 | MDQ6VXNlcjMyOTIwODgz | User | false |
openclaw/openclaw | 4,037,768,226 | I_kwDOQb6kR87wq3Qi | 38,671 | https://github.com/openclaw/openclaw/issues/38671 | https://api.github.com/repos/openclaw/openclaw/issues/38671 | [Feature]: Support direct Shell command execution in Cron to bypass LLM inference and context pollution | ### Summary
<h2 data-path-to-node="1" style="font-family: 'Google Sans', sans-serif !important; line-height: 1.15 !important; margin-top: 20px !important;">Problem to solve</h2>
<p data-path-to-node="2" style="font-family: 'Google Sans', sans-serif !important;">当前 OpenClaw 的 Cron 系统强制绑定 Agent 链路，导致三个核心痛点：</p>
<ul data... | closed | not_planned | true | 2 | [
"enhancement"
] | [] | 2026-03-07T07:38:09Z | 2026-04-29T04:34:41Z | 2026-04-26T18:56:58Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | fanzh365 | 14,065,918 | MDQ6VXNlcjE0MDY1OTE4 | User | false |
openclaw/openclaw | 4,037,941,213 | I_kwDOQb6kR87wrhfd | 38,702 | https://github.com/openclaw/openclaw/issues/38702 | https://api.github.com/repos/openclaw/openclaw/issues/38702 | [Feature]: WebChat input history navigation with ArrowUp / ArrowDown | ### Summary
Add shell-style input history recall to Control UI/WebChat:
- `ArrowUp` recalls previous sent input
- `ArrowDown` navigates forward
- returning to newest position restores current draft
### Problem to solve
In iterative workflows, users often resend or edit recent prompts.
WebChat currently has no keybo... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-03-07T08:54:00Z | 2026-04-30T04:34:39Z | 2026-04-27T11:08:57Z | CONTRIBUTOR | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Ivocin | 16,836,801 | MDQ6VXNlcjE2ODM2ODAx | User | false |
openclaw/openclaw | 4,038,621,818 | I_kwDOQb6kR87wuHp6 | 38,931 | https://github.com/openclaw/openclaw/issues/38931 | https://api.github.com/repos/openclaw/openclaw/issues/38931 | [Feature]: SSRF 内网访问确认机制 | ### Summary
# Feature Request: SSRF 内网访问确认机制
## 📋 问题描述
当前 OpenClaw 的 SSRF 策略只有两个选项：
- `dangerouslyAllowPrivateNetwork: true` - 直接允许访问内网
- `dangerouslyAllowPrivateNetwork: false` - 直接拒绝访问内网
**问题**：缺少"询问用户确认"的中间选项，导致：
1. 开启后 Agent 可能被恶意注入访问内网资源
2. 关闭后 Agent 无法在用户授权下访问内网设备（路由器、NAS、本地开发服务器等）
## 🎯 期望功能
### 方案 A：交互式确... | open | null | false | 2 | [
"enhancement"
] | [] | 2026-03-07T14:36:39Z | 2026-05-04T06:59:19Z | null | NONE | null | 20260504T090105Z | 2026-05-04T09:01:05Z | wwjj999 | 80,992,927 | MDQ6VXNlcjgwOTkyOTI3 | User | false |
openclaw/openclaw | 4,329,738,661 | I_kwDOQb6kR88AAAABAhKRpQ | 71,873 | https://github.com/openclaw/openclaw/issues/71873 | https://api.github.com/repos/openclaw/openclaw/issues/71873 | [Bug]: Gateway data loss: Silent read error on `paired.json` leads to overwrite of all pairings | ### Summary
A critical data loss bug exists in the OpenClaw Gateway's device and node pairing logic. If `paired.json` cannot be read on gateway startup for any transient reason (e.g., file lock, temporary I/O error, file corruption), the gateway proceeds with an empty in-memory list of pairings. The next time the stat... | closed | completed | true | 1 | [] | [] | 2026-04-26T01:26:34Z | 2026-04-28T04:43:15Z | 2026-04-26T01:51:07Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | iret77 | 63,622,643 | MDQ6VXNlcjYzNjIyNjQz | User | false |
openclaw/openclaw | 4,329,724,635 | I_kwDOQb6kR88AAAABAhJa2w | 71,869 | https://github.com/openclaw/openclaw/issues/71869 | https://api.github.com/repos/openclaw/openclaw/issues/71869 | [Bug]: Bonjour plugin does not wire Ciao cancellation rejection handler, causing unhandled rejection on CIAO PROBING CANCELLED | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
OpenClaw 2026.4.24 reports an unhandled promise rejection for `CIAO PROBING CANCELLED` while the bundled Bonjour plugin contains a Ciao cancellation classifier that is not wired into the advertised runtime call.
### Steps to ... | closed | completed | true | 2 | [
"bug",
"bug:crash"
] | [] | 2026-04-26T01:19:34Z | 2026-04-28T04:43:14Z | 2026-04-26T01:54:41Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | yoon1012 | 7,987,078 | MDQ6VXNlcjc5ODcwNzg= | User | false |
openclaw/openclaw | 3,994,717,806 | I_kwDOQb6kR87uGo5u | 27,445 | https://github.com/openclaw/openclaw/issues/27445 | https://api.github.com/repos/openclaw/openclaw/issues/27445 | [Feature]: `announceTarget` option for sub-agent completion announce routing | ### Summary
Allow sub-agent completion announces to route to the parent session as a user-message trigger instead of directly to the channel, enabling the main agent to orchestrate multi-step workflows.
### Problem to solve
When a sub-agent spawned via `sessions_spawn` completes, the completion announce is dispatche... | open | null | false | 8 | [
"enhancement"
] | [] | 2026-02-26T10:39:02Z | 2026-05-05T03:10:32Z | null | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | good-warning | 132,053,843 | U_kgDOB977Uw | User | false |
openclaw/openclaw | 3,995,006,324 | I_kwDOQb6kR87uHvV0 | 27,482 | https://github.com/openclaw/openclaw/issues/27482 | https://api.github.com/repos/openclaw/openclaw/issues/27482 | [Feature]: Support direct video upload to LLM | ### Summary
Enable direct video file upload to LLM models that support native video understanding, similar to existing image upload functionality.
### Problem to solve
Many modern LLMs (GPT-4V, Claude 3, Qwen-VL, etc.) now natively support video understanding. However, OpenClaw currently only supports image uploads,... | open | null | false | 2 | [
"enhancement"
] | [] | 2026-02-26T11:33:10Z | 2026-05-05T02:56:11Z | null | NONE | null | 20260505T030055Z | 2026-05-05T03:00:55Z | Levis-ghub | 42,949,472 | MDQ6VXNlcjQyOTQ5NDcy | User | false |
openclaw/openclaw | 3,995,351,426 | I_kwDOQb6kR87uJDmC | 27,526 | https://github.com/openclaw/openclaw/issues/27526 | https://api.github.com/repos/openclaw/openclaw/issues/27526 | [Feature]: [Feature Request] OpenClaw WebChat: Conversation Management & Multi-AI Collaboration | ### Summary
enhancement, webchat, feature-request, discussion
### Problem to solve
# [Feature Request] OpenClaw WebChat: Conversation Management & Multi-AI Collaboration
## Problem
1. **Poor conversation management**: All chats mixed, hard to find/switch between topics
2. **Single-AI limitation**: Missing multi-per... | open | null | false | 3 | [
"enhancement"
] | [] | 2026-02-26T12:49:01Z | 2026-05-05T03:10:33Z | null | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | jhfVPN | 151,948,295 | U_kgDOCQ6MBw | User | false |
openclaw/openclaw | 3,995,584,520 | I_kwDOQb6kR87uJ8gI | 27,574 | https://github.com/openclaw/openclaw/issues/27574 | https://api.github.com/repos/openclaw/openclaw/issues/27574 | [Feature]: Browser Preview Panel in Dashboard | ### Summary
Add a live browser preview panel to the OpenClaw Dashboard that shows what the built-in browser sees in real-time.
### Problem to solve
Use Case
Users running OpenClaw on a VPS (in a container/Docker) want to watch the AI browse the web and troubleshoot issues. Currently, there's no way to see what the ... | open | null | false | 2 | [
"enhancement"
] | [] | 2026-02-26T13:29:41Z | 2026-05-05T02:56:12Z | null | NONE | null | 20260505T030055Z | 2026-05-05T03:00:55Z | thatdudepaco888 | 264,172,196 | U_kgDOD77ypA | User | false |
openclaw/openclaw | 3,998,279,156 | I_kwDOQb6kR87uUOX0 | 28,025 | https://github.com/openclaw/openclaw/issues/28025 | https://api.github.com/repos/openclaw/openclaw/issues/28025 | [Feature]: Thread replies when agent uses tools; top-level for conversational responses | ### Summary
Add a config option (e.g. channels.slack.thread.replyOnToolUse: true) that automatically routes agent replies to a thread when the session invoked one or more tools, and keeps replies at top-level when the
response is purely conversational.
### Problem to solve
Currently (post-v2026.2.24) agents never ... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-02-26T23:30:24Z | 2026-05-05T03:10:35Z | null | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | seterhenakbar | 11,371,503 | MDQ6VXNlcjExMzcxNTAz | User | false |
openclaw/openclaw | 3,999,070,734 | I_kwDOQb6kR87uXPoO | 28,303 | https://github.com/openclaw/openclaw/issues/28303 | https://api.github.com/repos/openclaw/openclaw/issues/28303 | Rename i18n locale codes from region-based (zh-CN/zh-TW) to script-based (zh-Hans/zh-Hant) | ### Summary
Rename locale identifiers from region-based codes (zh-CN, zh-TW) to script-based codes (zh-Hans, zh-Hant) per BCP 47 best practices.
### Problem to solve
The current i18n system uses `zh-CN` and `zh-TW` as locale identifiers. These are region-based codes that technically mean "Chinese as used in China" ... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-02-27T04:49:47Z | 2026-05-05T02:56:17Z | null | NONE | null | 20260505T030055Z | 2026-05-05T03:00:55Z | xingzihai | 87,858,694 | MDQ6VXNlcjg3ODU4Njk0 | User | false |
openclaw/openclaw | 4,001,492,457 | I_kwDOQb6kR87uge3p | 28,847 | https://github.com/openclaw/openclaw/issues/28847 | https://api.github.com/repos/openclaw/openclaw/issues/28847 | Provider key pools + bucket-aware cooldown/backoff on 429 (Gemini projects) to prevent retry storms | ### Summary
Key manager for multiple api keys.
### Problem to solve
Problem
OpenClaw currently treats provider API keys as effectively static. Under concurrency (especially Gemini free tier), it’s easy to hit 429 RESOURCE_EXHAUSTED and then repeatedly retry the same key/project, causing long stalls, wasted tokens, ... | open | null | false | 2 | [
"enhancement"
] | [] | 2026-02-27T15:11:12Z | 2026-05-05T03:10:36Z | null | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | pyro0ownz | 40,363,536 | MDQ6VXNlcjQwMzYzNTM2 | User | false |
openclaw/openclaw | 4,001,723,260 | I_kwDOQb6kR87uhXN8 | 28,913 | https://github.com/openclaw/openclaw/issues/28913 | https://api.github.com/repos/openclaw/openclaw/issues/28913 | [Feature]: Source-based filtering for autoCapture | ### Summary
Add a captureFilter option to autoCapture that excludes content wrapped in security tags (e.g., EXTERNAL_UNTRUSTED_CONTENT) from being stored as memories.
### Problem to solve
When autoCapture: true is enabled, the shouldCapture() function uses regex triggers like “always”, “never”, “prefer”, and “rememb... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-02-27T16:00:38Z | 2026-05-05T02:56:20Z | null | CONTRIBUTOR | null | 20260505T030055Z | 2026-05-05T03:00:55Z | BKF-Gitty | 263,413,630 | U_kgDOD7Nffg | User | false |
openclaw/openclaw | 4,001,943,549 | I_kwDOQb6kR87uiM_9 | 28,965 | https://github.com/openclaw/openclaw/issues/28965 | https://api.github.com/repos/openclaw/openclaw/issues/28965 | [Proposal] Simplify Model Configuration: Single-Source Reconcile + Session Override Reset | ### Summary
目前 OpenClaw 的模型相关配置在实际运维中容易出现“需要改多个位置才生效”的问题，导致配置复杂、行为不透明、排障成本高。
希望和维护者讨论：是否可以引入“单一事实源 + 一键收敛”的方案，减少手动多点修改。
Today, model configuration in OpenClaw can require updates across multiple state/config locations before behavior fully matches user intent.
This increases operational complexity and makes model... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-02-27T16:54:10Z | 2026-05-05T03:10:37Z | null | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | KurohaneKaoruko | 101,451,942 | U_kgDOBgwIpg | User | false |
openclaw/openclaw | 4,002,626,231 | I_kwDOQb6kR87ukzq3 | 29,100 | https://github.com/openclaw/openclaw/issues/29100 | https://api.github.com/repos/openclaw/openclaw/issues/29100 | [Feature]: Cost metadata support for auto-discovered Amazon Bedrock models | ### Summary
Allow cost-per-token metadata to be attached to Bedrock models discovered via models.bedrockDiscovery without requiring a duplicate explicit provider block.
### Problem to solve
When using Amazon Bedrock with IAM role-based auto-discovery (models.bedrockDiscovery), there is no supported way to attach cos... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-02-27T19:54:59Z | 2026-05-05T02:56:21Z | null | NONE | null | 20260505T030055Z | 2026-05-05T03:00:55Z | mg-mavios | 163,306,920 | U_kgDOCbvdqA | User | false |
openclaw/openclaw | 4,002,824,854 | I_kwDOQb6kR87ulkKW | 29,132 | https://github.com/openclaw/openclaw/issues/29132 | https://api.github.com/repos/openclaw/openclaw/issues/29132 | [Feature]: Add pre-LLM tool result sanitization hook for deterministic policy enforcement | ### Summary
Introduce a new synchronous plugin hook (`tool_result_before_model`) that transforms tool results before any model step can consume them..
### Problem to solve
tool_result_persist is useful for transcript hygiene but is not an explicit pre-LLM security boundary. For dynamic security policies (e.g., strip... | open | null | false | 2 | [
"enhancement"
] | [] | 2026-02-27T20:53:58Z | 2026-05-05T03:10:39Z | null | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | DanMcInerney | 1,813,669 | MDQ6VXNlcjE4MTM2Njk= | User | false |
openclaw/openclaw | 4,003,229,581 | I_kwDOQb6kR87unG-N | 29,195 | https://github.com/openclaw/openclaw/issues/29195 | https://api.github.com/repos/openclaw/openclaw/issues/29195 | ACP + Codex: silent failures, stalled sessions, and missing permission docs | ## Summary
ACP sessions with Codex (`sessions_spawn` with `runtime: "acp"`, `agentId: "codex"`) have several issues that make them unreliable for non-trivial coding tasks.
## Issues encountered
### 1. Silent permission failure with no user-facing error
**Default `permissionMode` is `approve-reads`**, which causes C... | open | null | false | 1 | [] | [
"vincentkoc"
] | 2026-02-27T23:04:19Z | 2026-05-05T03:10:40Z | null | CONTRIBUTOR | null | 20260505T060034Z | 2026-05-05T06:00:34Z | barronlroth | 5,567,884 | MDQ6VXNlcjU1Njc4ODQ= | User | false |
openclaw/openclaw | 4,003,732,616 | I_kwDOQb6kR87upByI | 29,369 | https://github.com/openclaw/openclaw/issues/29369 | https://api.github.com/repos/openclaw/openclaw/issues/29369 | FEATURE: Add tools parameter to sessions_spawn | ### Summary
FEATURE — Spawn Tools Parameter
───
Allow passing a tool allowlist directly in sessions_spawn calls, so sub-agent tool permissions can be configured at spawn time without requiring separate agent profiles.
### Problem to solve
Use Case
Currently, the only way to give different sub-agents diffe... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-02-28T03:08:04Z | 2026-05-04T04:37:03Z | 2026-05-01T09:43:38Z | NONE | null | 20260504T060031Z | 2026-05-04T06:00:31Z | kAIborg24 | 261,855,322 | U_kgDOD5uYWg | User | false |
openclaw/openclaw | 4,003,734,978 | I_kwDOQb6kR87upCXC | 29,372 | https://github.com/openclaw/openclaw/issues/29372 | https://api.github.com/repos/openclaw/openclaw/issues/29372 | FEATURE: Per-agent subagents.workspace config | ### Summary
FEATURE — Subagent Workspace Override
───
Add an agents.list[].subagents.workspace config option that specifies the workspace directory for child sub-agents spawned by that agent. This would allow parent agents to isolate their default sub-agents without requiring each sub-agent to have its own agent pro... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-02-28T03:09:53Z | 2026-05-05T02:56:23Z | null | NONE | null | 20260505T030055Z | 2026-05-05T03:00:55Z | kAIborg24 | 261,855,322 | U_kgDOD5uYWg | User | false |
openclaw/openclaw | 4,003,761,051 | I_kwDOQb6kR87upIub | 29,384 | https://github.com/openclaw/openclaw/issues/29384 | https://api.github.com/repos/openclaw/openclaw/issues/29384 | [Feature]: Support for sandbox Docker parameters (--gpus all, --ipc=host) | ### Summary
Allow passing custom Docker parameters like `--gpus all` and `--ipc=host` to the sandbox environment.
### Problem to solve
Currently, the sandbox environment lacks the ability to leverage GPUs or host IPC, which is necessary for running GPU-accelerated tasks.
### Proposed solution
Add configuration opt... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-02-28T03:27:27Z | 2026-05-05T02:56:24Z | null | NONE | null | 20260505T030055Z | 2026-05-05T03:00:55Z | panyz522 | 6,896,240 | MDQ6VXNlcjY4OTYyNDA= | User | false |
openclaw/openclaw | 4,003,979,985 | I_kwDOQb6kR87up-LR | 29,487 | https://github.com/openclaw/openclaw/issues/29487 | https://api.github.com/repos/openclaw/openclaw/issues/29487 | [Feature]: [Feishu] Include quoted/parent message content when forwarding events to agent | ### Summary
[Feishu] Include quoted/parent message content when forwarding events to agent
### Problem to solve
When a user quotes a message in Feishu and @mentions the bot, the agent only receives the new message text. The quoted (parent) message content is not included in the context, even though parent_id is pres... | open | null | false | 2 | [
"enhancement"
] | [] | 2026-02-28T05:49:56Z | 2026-05-05T03:10:43Z | null | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | maxinezhangmei-bot | 248,855,549 | U_kgDODtU7_Q | User | false |
openclaw/openclaw | 4,004,493,563 | I_kwDOQb6kR87ur7j7 | 29,707 | https://github.com/openclaw/openclaw/issues/29707 | https://api.github.com/repos/openclaw/openclaw/issues/29707 | [Feature Request]: Native Prompt Caching Support for Amazon Nova (Bedrock) | ### Summary
Requesting the removal of the hardcoded cacheRetention: "none" restriction for Amazon Nova models on Bedrock. Now that Bedrock supports native caching for Nova 2 Lite, Nova Pro, Lite, and Micro, OpenClaw should allow these models to utilize caching to reduce costs by up to 90% and latency by 85%.
### Prob... | open | null | false | 2 | [
"enhancement"
] | [] | 2026-02-28T11:25:00Z | 2026-05-05T03:10:44Z | null | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | tribecar-shurn | 119,289,680 | U_kgDOBxw3UA | User | false |
openclaw/openclaw | 4,004,529,433 | I_kwDOQb6kR87usEUZ | 29,725 | https://github.com/openclaw/openclaw/issues/29725 | https://api.github.com/repos/openclaw/openclaw/issues/29725 | [Feature]: Set Discord Presence | ### Summary
Add a `updatePresence()` method to `api.runtime.channel.discord` so plugins can dynamically update the bot's Discord activity and status at runtime.
### Problem to solve
Plugins have no supported way to update Discord presence dynamically. The Discord client is owned by the bundled channel implementation... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-02-28T11:52:49Z | 2026-05-05T03:10:46Z | null | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | fedtoolittle | 33,679,616 | MDQ6VXNlcjMzNjc5NjE2 | User | false |
openclaw/openclaw | 4,004,944,934 | I_kwDOQb6kR87utpwm | 29,907 | https://github.com/openclaw/openclaw/issues/29907 | https://api.github.com/repos/openclaw/openclaw/issues/29907 | Feature: New cron kind `exec`/`lobster` — run pipelines without LLM, trigger agent only on condition | ## Problem
All cron jobs today use `kind: "agentTurn"`, which always spins up a full LLM session — system prompt (~25K tokens), tool definitions, context loading, etc. — even for completely deterministic tasks.
**Real-world example:** We built a Lobster pipeline for n8n health monitoring that uses `diff.last` to dete... | open | null | false | 1 | [] | [
"vignesh07"
] | 2026-02-28T16:34:35Z | 2026-05-05T03:10:47Z | null | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | alexph-dev | 129,837,470 | U_kgDOB70png | User | false |
openclaw/openclaw | 4,005,035,238 | I_kwDOQb6kR87ut_zm | 29,945 | https://github.com/openclaw/openclaw/issues/29945 | https://api.github.com/repos/openclaw/openclaw/issues/29945 | Security: unpaired message response leaks platform identity | ## Problem
When an unpaired user messages on WhatsApp/Discord/etc, OpenClaw responds with a branded message mentioning OpenClaw + a pairing code. This confirms the platform to potential attackers and gives them a fingerprint for targeted social engineering.
## Suggestion
Default to **silent mode** — don't respond at... | open | null | false | 3 | [] | [
"vincentkoc"
] | 2026-02-28T17:26:43Z | 2026-05-05T02:56:29Z | null | NONE | null | 20260505T030055Z | 2026-05-05T03:00:55Z | andrewchen | 82,151 | MDQ6VXNlcjgyMTUx | User | false |
openclaw/openclaw | 4,005,376,990 | I_kwDOQb6kR87uvTPe | 30,067 | https://github.com/openclaw/openclaw/issues/30067 | https://api.github.com/repos/openclaw/openclaw/issues/30067 | [Feature]: Initial Setup Heads-Up Notification | ### Summary
Feature Request: Initial Setup Heads-Up Notification
Description: ... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-02-28T20:52:23Z | 2026-05-05T03:10:49Z | null | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | roclob | 264,691,270 | U_kgDOD8beRg | User | false |
openclaw/openclaw | 4,006,479,094 | I_kwDOQb6kR87uzgT2 | 30,389 | https://github.com/openclaw/openclaw/issues/30389 | https://api.github.com/repos/openclaw/openclaw/issues/30389 | [Feature Request] Add file upload button and model/skill selection menu to WebChat UI | ### Summary
Enhance WebChat UI with file upload button, model selection dropdown, and skill/capability menu for better user experience.
### Problem to solve
Currently, the WebChat UI lacks several basic features that limit its usability compared to other channels like Feishu, Telegram, and Discord:
1. **No file upl... | closed | completed | false | 1 | [
"enhancement"
] | [] | 2026-03-01T06:42:08Z | 2026-05-05T03:19:53Z | 2026-05-05T03:19:53Z | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | canmaxfire | 76,983,334 | MDQ6VXNlcjc2OTgzMzM0 | User | false |
openclaw/openclaw | 4,006,642,089 | I_kwDOQb6kR87u0IGp | 30,452 | https://github.com/openclaw/openclaw/issues/30452 | https://api.github.com/repos/openclaw/openclaw/issues/30452 | [Feature]: Compaction: configurable model override + flush-then-reset mode | ### Summary
Compaction: configurable model override + flush-then-reset mode
### Problem to solve
### Problem
Compaction and memory flush both use the session model with no override. For Opus sessions ($5/$25 per MTok), housekeeping costs dominate:
- **Compaction (default mode):** ~$0.90-1.13 per cycle (100k histor... | open | null | false | 3 | [
"enhancement"
] | [] | 2026-03-01T08:20:37Z | 2026-05-05T03:10:56Z | null | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | dzhang253 | 261,981,134 | U_kgDOD52Dzg | User | false |
openclaw/openclaw | 4,006,770,303 | I_kwDOQb6kR87u0nZ_ | 30,518 | https://github.com/openclaw/openclaw/issues/30518 | https://api.github.com/repos/openclaw/openclaw/issues/30518 | [Feature]: pin/unpin message action for Telegram | ### Summary
The message tool supports delete, edit, react, send, sticker for Telegram but has no pin/unpin actions.
Telegram Bot API supports pinChatMessage and unpinChatMessage — would be great to expose as pin/unpin actions.
Use case: Agents maintaining pinned status dashboards need to pin new messages programmati... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-03-01T09:39:49Z | 2026-05-05T03:10:57Z | null | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | dan-osss | 261,756,476 | U_kgDOD5oWPA | User | false |
openclaw/openclaw | 4,006,803,342 | I_kwDOQb6kR87u0veO | 30,529 | https://github.com/openclaw/openclaw/issues/30529 | https://api.github.com/repos/openclaw/openclaw/issues/30529 | [Feature]: Support typing indicators for Discord in message tool | ### Summary
## Feature request: Discord typing indicators in message tool
... | open | null | false | 4 | [
"enhancement"
] | [] | 2026-03-01T09:58:23Z | 2026-05-05T03:10:59Z | null | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | sgarc57 | 22,865,328 | MDQ6VXNlcjIyODY1MzI4 | User | false |
openclaw/openclaw | 4,007,523,419 | I_kwDOQb6kR87u3fRb | 30,759 | https://github.com/openclaw/openclaw/issues/30759 | https://api.github.com/repos/openclaw/openclaw/issues/30759 | [Feature]: WebChat: Add file upload support for documents (PDF, Word, TXT) | ### Summary
Add file upload capability to WebChat channel, allowing users to send PDF, Word, and TXT files directly in the web interface.
## Motivation
Currently, WebChat users cannot share documents with the AI assistant. This forces users to:
- Copy-paste long text manually
- Switch to other channels (Telegram, Dis... | closed | completed | false | 1 | [
"enhancement"
] | [] | 2026-03-01T15:40:03Z | 2026-05-03T21:24:11Z | 2026-05-03T21:24:11Z | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | uhang | 18,619,189 | MDQ6VXNlcjE4NjE5MTg5 | User | false |
openclaw/openclaw | 4,007,751,930 | I_kwDOQb6kR87u4XD6 | 30,828 | https://github.com/openclaw/openclaw/issues/30828 | https://api.github.com/repos/openclaw/openclaw/issues/30828 | [Feature]: Rate limit models by tokens per interval and requests per interval | ### Summary
I would like to be able to manually set:
1. Tokens per N seconds/minutes/hours/...
2. Requests per N seconds/minutes/hours/...
Correct me if I am wrong, but this isn't implemeted yet, nor is it mentioned in DOCS.
### Problem to solve
I hit rate limit issue quite often.
### Proposed solution
Rate limit... | open | null | false | 2 | [
"enhancement"
] | [] | 2026-03-01T17:10:34Z | 2026-05-05T02:56:32Z | null | NONE | null | 20260505T030055Z | 2026-05-05T03:00:55Z | manfromfms | 61,939,162 | MDQ6VXNlcjYxOTM5MTYy | User | false |
openclaw/openclaw | 4,007,876,732 | I_kwDOQb6kR87u41h8 | 30,861 | https://github.com/openclaw/openclaw/issues/30861 | https://api.github.com/repos/openclaw/openclaw/issues/30861 | [Feature]: Persistent active-agent + workspace indicator in Dashboard (multi-agent safety) | ### Summary
Show the currently active agent and workspace persistently in the web dashboard, with clear switch feedback, to avoid wrong-agent interactions in multi-agent workflows.
### Problem to solve
In multi-agent setups (e.g. main + codex), it is not explicit enough which agent/session is currently active in the... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-03-01T18:04:19Z | 2026-05-05T02:56:33Z | null | NONE | null | 20260505T030055Z | 2026-05-05T03:00:55Z | marcuspocus | 236,603 | MDQ6VXNlcjIzNjYwMw== | User | false |
openclaw/openclaw | 4,009,589,950 | I_kwDOQb6kR87u_Xy- | 31,379 | https://github.com/openclaw/openclaw/issues/31379 | https://api.github.com/repos/openclaw/openclaw/issues/31379 | [Feature]: Channels settings page should always show built-in channels alongside custom channels | ### Summary
When a custom channel plugin (e.g. `icenter`) is enabled via `plugins.entries`, the Channels settings page in the Control UI only displays that custom channel. All built-in channels (WhatsApp, Telegram, Discord, Google Chat, Slack, Signal, iMessage, Nostr) disappear from the page entirely.
### Problem to... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-03-02T06:40:00Z | 2026-05-05T03:11:00Z | null | CONTRIBUTOR | null | 20260505T060034Z | 2026-05-05T06:00:34Z | BUGKillerKing | 117,326,392 | U_kgDOBv5COA | User | false |
openclaw/openclaw | 4,009,670,701 | I_kwDOQb6kR87u_rgt | 31,396 | https://github.com/openclaw/openclaw/issues/31396 | https://api.github.com/repos/openclaw/openclaw/issues/31396 | [Feature]: Whenever I ask him to switch models or connect to a new model on his own, it often freezes. How can I avoid this? | ### Summary
Whenever I ask him to switch models or connect to a new model on his own, it often freezes. How can I avoid this?
### Problem to solve
At least don't let yourself crash when switching models, please.
### Proposed solution
Transaction isolation or sandbox?
### Alternatives considered
_No response_
##... | closed | not_planned | true | 2 | [
"enhancement"
] | [] | 2026-03-02T07:05:05Z | 2026-05-04T04:36:08Z | 2026-05-01T14:51:25Z | NONE | null | 20260504T060031Z | 2026-05-04T06:00:31Z | Fission21 | 80,576,421 | MDQ6VXNlcjgwNTc2NDIx | User | false |
openclaw/openclaw | 4,010,504,560 | I_kwDOQb6kR87vC3Fw | 31,584 | https://github.com/openclaw/openclaw/issues/31584 | https://api.github.com/repos/openclaw/openclaw/issues/31584 | [Feature]: OpenClaw non-interactive setup for GitHub Copilot | ### Summary
OpenClaw supports two ways to onboard to Copilot - using GH copilot provider (interactive token flow) or VS code proxy plugin.
<img width="628" height="111" alt="Image" src="https://github.com/user-attachments/assets/79ac9d15-b505-4f1a-ba8a-6184adf38c66" />
https://docs.openclaw.ai/providers/github-copilo... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-02T10:32:51Z | 2026-04-29T04:34:06Z | 2026-04-26T21:30:31Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | shubham1172 | 22,132,944 | MDQ6VXNlcjIyMTMyOTQ0 | User | false |
openclaw/openclaw | 4,013,385,151 | I_kwDOQb6kR87vN2W_ | 32,188 | https://github.com/openclaw/openclaw/issues/32188 | https://api.github.com/repos/openclaw/openclaw/issues/32188 | [Feature]: | ### Summary
There are workarounds, but I'd like to see a status area which would display which agent is being used at any given time, and perhaps some note about tokens remaining or no tokens required or. Something like that, right in the interface. As it is, there's no easy way to see that information other than to... | open | null | false | 2 | [
"enhancement"
] | [] | 2026-03-02T21:27:48Z | 2026-05-05T02:56:41Z | null | NONE | null | 20260505T030055Z | 2026-05-05T03:00:55Z | dmccolly | 128,931,847 | U_kgDOB69YBw | User | false |
openclaw/openclaw | 4,014,347,401 | I_kwDOQb6kR87vRhSJ | 32,438 | https://github.com/openclaw/openclaw/issues/32438 | https://api.github.com/repos/openclaw/openclaw/issues/32438 | [Feature]: Add a formatted log file when JSON logs are hard to read and Control UI is unusable | ### Summary
When file logging is enabled, the logger writes a second formatted log file next to the existing JSON log (e.g. openclaw.log → openclaw-formatted.log). Each line is plain text: YYYY-MM-DD HH:mm:ss [level] [subsystem] message. The original JSON log file is unchanged; CLI, Control UI, and scripts keep using ... | open | null | false | 4 | [
"enhancement"
] | [] | 2026-03-03T02:35:51Z | 2026-05-05T02:56:44Z | null | CONTRIBUTOR | null | 20260505T030055Z | 2026-05-05T03:00:55Z | maweibin | 18,023,423 | MDQ6VXNlcjE4MDIzNDIz | User | false |
openclaw/openclaw | 4,014,634,401 | I_kwDOQb6kR87vSnWh | 32,530 | https://github.com/openclaw/openclaw/issues/32530 | https://api.github.com/repos/openclaw/openclaw/issues/32530 | [Feature]: Auto-discovery of agent configurations from external workspaces | ### Summary
Enable OpenClaw to automatically discover and load agent configurations from specified workspace directories, eliminating the need for manual agent registration in the central configuration file.
### Problem to solve
Currently, OpenClaw requires all agents to be explicitly defined in ~/.openclaw/openclaw... | open | null | false | 2 | [
"enhancement"
] | [] | 2026-03-03T04:26:08Z | 2026-05-05T02:56:48Z | null | NONE | null | 20260505T030055Z | 2026-05-05T03:00:55Z | veictry | 20,666,224 | MDQ6VXNlcjIwNjY2MjI0 | User | false |
openclaw/openclaw | 4,016,089,171 | I_kwDOQb6kR87vYKhT | 32,932 | https://github.com/openclaw/openclaw/issues/32932 | https://api.github.com/repos/openclaw/openclaw/issues/32932 | [Feature]: Suggestion for scheduled task, WebHook URL function | ### Summary
# Suggestion for cron job, WebHook URL function
The original intention of the Webhook URL function is to send the results of scheduled task execution to a specified URL endpoint
<img width="1440" height="692" alt="Image" src="https://github.com/user-attachments/assets/2f050116-3bd0-49db-be96-7e79cfa354... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-03T11:12:23Z | 2026-04-29T04:34:12Z | 2026-04-26T19:53:15Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | KaGty1 | 153,613,891 | U_kgDOCSf2Qw | User | false |
openclaw/openclaw | 4,019,236,208 | I_kwDOQb6kR87vkK1w | 33,624 | https://github.com/openclaw/openclaw/issues/33624 | https://api.github.com/repos/openclaw/openclaw/issues/33624 | [Feature]: Expose Swagger open api schema | ### Summary
It would be nice to have open api schema to interact with open claw
### Problem to solve
I would like to generate custom api client to interact with open claw programmatically
### Proposed solution
Generate open api schema from existing andpoints as a part of opensource repository
### Alternatives ... | open | null | false | 2 | [
"enhancement"
] | [] | 2026-03-03T23:35:50Z | 2026-05-05T02:56:56Z | null | NONE | null | 20260505T030055Z | 2026-05-05T03:00:55Z | vasilich6107 | 6,721,286 | MDQ6VXNlcjY3MjEyODY= | User | false |
openclaw/openclaw | 4,019,760,959 | I_kwDOQb6kR87vmK8_ | 33,832 | https://github.com/openclaw/openclaw/issues/33832 | https://api.github.com/repos/openclaw/openclaw/issues/33832 | [Feature] Gateway: Log request/response timing and size for model API calls | ### Summary
Add Gateway logs for model API calls showing request/response timing and body size.
like
[10:32:15.123] → Model API request: 12.5 KB (~3200 tokens), model: gpt-5.2
[10:32:18.456] ← Model API response: 2.3 KB, latency: 3.33s
### Problem to solve
When debugging slow AI responses, I can't tell where the del... | closed | completed | true | 4 | [
"enhancement"
] | [
"vincentkoc"
] | 2026-03-04T03:04:07Z | 2026-04-29T04:34:10Z | 2026-04-26T20:44:37Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | wwh830 | 96,802,767 | U_kgDOBcUXzw | User | false |
openclaw/openclaw | 4,019,786,702 | I_kwDOQb6kR87vmRPO | 33,838 | https://github.com/openclaw/openclaw/issues/33838 | https://api.github.com/repos/openclaw/openclaw/issues/33838 | [Feature]: need a mechanism for nodes to declare their command schemas | ### Summary
When a node registers custom commands to the gateway after pairing, the gateway has no knowledge of the command's input/output schema. We need
a mechanism for nodes to declare their command schemas (e.g., JSON Schema or similar) so the gateway can validate inputs, route correctly,
and expose proper AP... | closed | not_planned | false | 2 | [
"enhancement"
] | [] | 2026-03-04T03:15:06Z | 2026-05-03T18:51:19Z | 2026-05-03T18:51:19Z | NONE | null | 20260503T210034Z | 2026-05-03T21:00:34Z | MapleAura | 149,750,728 | U_kgDOCO0DyA | User | false |
openclaw/openclaw | 4,329,789,103 | I_kwDOQb6kR88AAAABAhNWrw | 71,888 | https://github.com/openclaw/openclaw/issues/71888 | https://api.github.com/repos/openclaw/openclaw/issues/71888 | Feature request: supported route for ChatGPT Pro / Deep Research style agents | It would be really useful to have a supported OpenClaw route for ChatGPT Pro style deep research / Research Pro workflows.
Right now I can use `openai-codex/gpt-5.5` successfully through the Codex/OpenAI subscription path, but there does not seem to be a valid model route for a higher tier research mode. I tried confi... | closed | completed | true | 1 | [] | [] | 2026-04-26T01:56:58Z | 2026-04-29T04:36:19Z | 2026-04-26T12:07:59Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | marcbal77 | 8,789,169 | MDQ6VXNlcjg3ODkxNjk= | User | false |
openclaw/openclaw | 4,025,184,678 | I_kwDOQb6kR87v63Gm | 35,203 | https://github.com/openclaw/openclaw/issues/35203 | https://api.github.com/repos/openclaw/openclaw/issues/35203 | [RFC] Multi-Agent Collaboration Enhancement: Capability Profiling + Shared Blackboard + Layered Memory + Token Cost Governance | ### Summary
Enhance OpenClaw's multi-agent system with capability profiling, shared blackboard, layered memory boundaries, and token cost governance to address information silos, ambiguous task delegation, and uncontrolled token consumption.
### Problem to solve
OpenClaw's current multi-agent architecture follows a ... | open | null | false | 7 | [
"enhancement"
] | [] | 2026-03-05T01:53:27Z | 2026-05-05T03:11:12Z | null | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | zz0116 | 16,729,173 | MDQ6VXNlcjE2NzI5MTcz | User | false |
openclaw/openclaw | 4,003,890,387 | I_kwDOQb6kR87upoTT | 29,442 | https://github.com/openclaw/openclaw/issues/29442 | https://api.github.com/repos/openclaw/openclaw/issues/29442 | [Security/Architecture/Feature] Preventing Prompt Injection in Cron Jobs via Privilege Separation & Tagging | ### Summary
Implement a Dual-LLM architecture that isolates untrusted data into a read-only worker model to prevent prompt injection and privilege escalation in cron jobs.
### Problem to solve
Right now, giving an agent autonomous cron capabilities in OpenClaw is basically handing over the keys to the kingdom and ho... | open | null | false | 3 | [
"enhancement"
] | [] | 2026-02-28T04:47:25Z | 2026-05-05T03:10:42Z | null | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | yasinldev | 90,653,146 | MDQ6VXNlcjkwNjUzMTQ2 | User | false |
openclaw/openclaw | 4,039,217,419 | I_kwDOQb6kR87wwZEL | 39,107 | https://github.com/openclaw/openclaw/issues/39107 | https://api.github.com/repos/openclaw/openclaw/issues/39107 | [Feature]: add interface on CLI and nodejs/python. so that openclaw can leverage itself to give some AI completions. | ### Summary
e.g.
If we are building some skill called "Excel data categorize skill" and in the skill: most of the time just compose a script, we eval that script to categorize rows'data. As we have to consider max token length then so usual method is to implement a for-loop and call AI completion during each loop iter... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-03-07T19:08:28Z | 2026-04-28T04:43:13Z | 2026-04-26T01:58:51Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | AntiMoron | 6,587,734 | MDQ6VXNlcjY1ODc3MzQ= | User | false |
openclaw/openclaw | 4,039,252,815 | I_kwDOQb6kR87wwhtP | 39,120 | https://github.com/openclaw/openclaw/issues/39120 | https://api.github.com/repos/openclaw/openclaw/issues/39120 | [Feature]: Support arbitrary workspace files in bootstrap context injection | ### Summary
The bootstrap-extra-files hook only accepts recognized basenames (AGENTS.md, SOUL.md, TOOLS.md, etc.). There's no way to auto-inject custom workspace files into session context at bootstrap time. This forces users to either embed custom context into AGENTS.md (mixing concerns) or spend a tool call reading ... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-03-07T19:24:24Z | 2026-05-04T06:59:58Z | null | NONE | null | 20260504T090105Z | 2026-05-04T09:01:05Z | drewmarq | 24,689,714 | MDQ6VXNlcjI0Njg5NzE0 | User | false |
openclaw/openclaw | 4,039,265,907 | I_kwDOQb6kR87wwk5z | 39,127 | https://github.com/openclaw/openclaw/issues/39127 | https://api.github.com/repos/openclaw/openclaw/issues/39127 | [Feature]: Per-session activity state (busy/idle + awaiting_user, awaiting_subagent) via gateway API + WS statechange | ### Summary
Add a gateway-owned, per-session activity system (sessions.activity.get/list plus WS statechange events) so any client can reliably show session running state with elapsed time and explicit waiting states (awaiting user, awaiting subagent), without polling and without guesswork.
### Problem to solve
Open... | open | null | false | 2 | [
"enhancement"
] | [
"vincentkoc"
] | 2026-03-07T19:30:24Z | 2026-05-04T07:00:17Z | null | NONE | null | 20260504T090105Z | 2026-05-04T09:01:05Z | sab-xan | 1,909,071 | MDQ6VXNlcjE5MDkwNzE= | User | false |
openclaw/openclaw | 4,039,319,526 | I_kwDOQb6kR87wwx_m | 39,141 | https://github.com/openclaw/openclaw/issues/39141 | https://api.github.com/repos/openclaw/openclaw/issues/39141 | [Feature]: Optional session activity watchdog and alerts (blocked by #39127) | ### Summary
Blocked by #39127
https://github.com/openclaw/openclaw/issues/39127
Add an optional gateway-side watchdog that consumes sessions.activity state and emits alerts (stale runs, waiting too long, child sessions stuck), so operators don’t have to build their own monitoring. Baseline should require no LLM toke... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-03-07T19:52:02Z | 2026-05-04T07:00:19Z | null | NONE | null | 20260504T090105Z | 2026-05-04T09:01:05Z | sab-xan | 1,909,071 | MDQ6VXNlcjE5MDkwNzE= | User | false |
openclaw/openclaw | 4,040,341,678 | I_kwDOQb6kR87w0riu | 39,421 | https://github.com/openclaw/openclaw/issues/39421 | https://api.github.com/repos/openclaw/openclaw/issues/39421 | [Feature]: Bot is lagging with communication in Slack Channel | ### Summary
In Slack we can chat directly with bot, and also tag Bot in the Channel.
But my experience with Bot in channel is very lagging, mostly Bot not answering my question.
Only If I remind directly to Bot then they will get back to me in Channel.
### Problem to solve
Expect behavior: Bot can reply in Slack C... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-03-08T03:38:07Z | 2026-04-28T04:43:10Z | 2026-04-26T02:00:44Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | piggybit | 55,419,584 | MDQ6VXNlcjU1NDE5NTg0 | User | false |
openclaw/openclaw | 4,040,448,474 | I_kwDOQb6kR87w1Fna | 39,457 | https://github.com/openclaw/openclaw/issues/39457 | https://api.github.com/repos/openclaw/openclaw/issues/39457 | [Feature]: post-update hook / lifecycle script support | ### Summary
Add support for a post-update lifecycle hook (e.g. ~/.openclaw/hooks/post-update.sh) that runs automatically after a successful update.
### Problem to solve
Running `openclaw update` overwrites the LaunchAgent plist, wiping any custom `ProgramArguments` or environment variable injections. Users who rely ... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-03-08T04:28:19Z | 2026-05-01T04:40:37Z | 2026-04-29T03:21:24Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | 2nd-Runner | 52,268,561 | MDQ6VXNlcjUyMjY4NTYx | User | false |
openclaw/openclaw | 4,040,724,750 | I_kwDOQb6kR87w2JEO | 39,579 | https://github.com/openclaw/openclaw/issues/39579 | https://api.github.com/repos/openclaw/openclaw/issues/39579 | [Feature]: Support setting userTimezone per agent | ### Summary
Add the ability to configure userTimezone on a per-agent basis in agents.list[].userTimezone.
### Problem to solve
Users with multiple agents handling different regions/accounts currently have to share a single global `agents.defaults.userTimezone`. This limits the ability for different agents to have ac... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-08T07:11:40Z | 2026-04-28T04:43:07Z | 2026-04-26T02:02:06Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | FMCorz | 240,112 | MDQ6VXNlcjI0MDExMg== | User | false |
openclaw/openclaw | 4,040,856,196 | I_kwDOQb6kR87w2pKE | 39,635 | https://github.com/openclaw/openclaw/issues/39635 | https://api.github.com/repos/openclaw/openclaw/issues/39635 | [Feature]: Support custom auto-injected bootstrap files via config (extraBootstrapFiles) | ### Summary
Allow users to specify additional workspace files for auto-injection at session start via `agents.defaults.extraBootstrapFiles` config.
### Problem to solve
The set of auto-injected bootstrap files is hardcoded: `AGENTS.md`, `SOUL.md`, `TOOLS.md`, `IDENTITY.md`, `USER.md`, `HEARTBEAT.md`, `MEMORY.md`. Th... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-08T08:37:18Z | 2026-04-29T04:34:40Z | 2026-04-26T18:57:02Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | orsi-bit | 75,600,835 | MDQ6VXNlcjc1NjAwODM1 | User | false |
openclaw/openclaw | 4,040,916,007 | I_kwDOQb6kR87w23wn | 39,659 | https://github.com/openclaw/openclaw/issues/39659 | https://api.github.com/repos/openclaw/openclaw/issues/39659 | [Feature]: First-class multi-instance management for Docker installs | ### Summary
### Overview
OpenClaw’s Docker install path works well for running a single instance, but having multiple instances running on a single host is still quite manual.
So I think this is a real product gap for power users and self-hosters.
### Why upstream
There are already community wrappers that prove this... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-03-08T09:17:23Z | 2026-05-04T09:03:31Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | diligentapple | 79,948,585 | MDQ6VXNlcjc5OTQ4NTg1 | User | false |
openclaw/openclaw | 4,040,959,527 | I_kwDOQb6kR87w3CYn | 39,685 | https://github.com/openclaw/openclaw/issues/39685 | https://api.github.com/repos/openclaw/openclaw/issues/39685 | [Feature]: Network Access Control (allowedDomains / denyDomains) — Egress Firewall | ### Summary
Add network-level access control (`allowedDomains`/`denyDomains`) to OpenClaw config. This allows users to restrict which external domains the agent's web tools (`web_fetch`, `web_search`) can access, utilizing and extending the existing `SsrFPolicy` infrastructure.
### Problem to solve
Currently, OpenCl... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-03-08T09:44:19Z | 2026-05-04T09:03:35Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | Ar3ss12 | 150,044,748 | U_kgDOCPGATA | User | false |
openclaw/openclaw | 4,041,176,772 | I_kwDOQb6kR87w33bE | 39,805 | https://github.com/openclaw/openclaw/issues/39805 | https://api.github.com/repos/openclaw/openclaw/issues/39805 | [Feature]: 飞书插件支持读取飞书表格 (Sheet) 内容 | ### Summary
## 问题描述
飞书官方插件 (@openclaw/feishu) 目前无法读取飞书表格 (Sheet) 的内容。
当用户发送飞书表格链接时（格式如 `https://xxx.feishu.cn/wiki/TOKEN`，实际类型为 sheet），插件无法读取表格数据。
## 当前行为
- 使用 `feishu_doc` 工具读取 → 返回 404（因为这是 Sheet 不是 Docx）
- 使用 `feishu_wiki` 工具 → 能识别节点类型是 sheet，但无法读取内容
- `feishu_bitable_*` 工具 → 仅支持多维表格 (Bitable)，不支持普通表格 (Sheet)
... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-08T12:12:26Z | 2026-04-29T04:36:41Z | 2026-04-26T11:53:13Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | weixiaojingwork-sketch | 265,690,300 | U_kgDOD9YcvA | User | false |
openclaw/openclaw | 4,041,206,573 | I_kwDOQb6kR87w3-st | 39,831 | https://github.com/openclaw/openclaw/issues/39831 | https://api.github.com/repos/openclaw/openclaw/issues/39831 | [Feature]: Display Raw JSON change lines in UI | ### Summary
One-line statement of the requested capability:
Show highlighted or annotated lines indicating changes in raw JSON data within the UI for easier debugging and review.
### Problem to solve
The current UI displays raw JSON without any visual differentiation for added, removed, or modified lines. When revie... | closed | not_planned | true | 3 | [
"enhancement"
] | [] | 2026-03-08T12:32:14Z | 2026-04-28T04:30:45Z | 2026-04-26T02:04:38Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | JiajunBernoulli | 45,968,640 | MDQ6VXNlcjQ1OTY4NjQw | User | false |
openclaw/openclaw | 4,041,311,446 | I_kwDOQb6kR87w4YTW | 39,879 | https://github.com/openclaw/openclaw/issues/39879 | https://api.github.com/repos/openclaw/openclaw/issues/39879 | [Feature] WhatsApp outbound mentions (mentionedJids) support | ### Summary
When the agent sends messages to WhatsApp groups using `@phonenumber`
in the text body, it doesn't trigger WhatsApp's native mention
notification — because the Baileys `mentionedJids` array isn't being
populated in the outbound payload.
### Problem to solve
**Problem**
When sending messages to WhatsAp... | closed | completed | false | 2 | [
"enhancement"
] | [] | 2026-03-08T13:31:59Z | 2026-05-03T04:50:56Z | 2026-05-03T04:50:56Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | kengi1437 | 19,283,996 | MDQ6VXNlcjE5MjgzOTk2 | User | false |
openclaw/openclaw | 4,041,329,400 | I_kwDOQb6kR87w4cr4 | 39,890 | https://github.com/openclaw/openclaw/issues/39890 | https://api.github.com/repos/openclaw/openclaw/issues/39890 | [Feature]: Restore recursive skill scanning for better skill organization | ### Summary
https://github.com/openclaw/openclaw/issues/38231
The issue page got an issue on displaying. So I reopen a new one here.
### Problem to solve
Currently, OpenClaw (v2026.3.7) only seems to scan the first level of the ~/.openclaw/skills/ directory. This makes it difficult to organize skills, especially fo... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-08T13:43:51Z | 2026-04-28T04:43:03Z | 2026-04-26T02:05:02Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | TimLai666 | 43,640,816 | MDQ6VXNlcjQzNjQwODE2 | User | false |
openclaw/openclaw | 4,041,486,173 | I_kwDOQb6kR87w5C9d | 39,979 | https://github.com/openclaw/openclaw/issues/39979 | https://api.github.com/repos/openclaw/openclaw/issues/39979 | [Feature]: Path-scoped RWX permissions for exec and file tools | ### Summary
Replace the binary-level exec allowlist with a path-keyed RWX permission map so agents
inherit read/write/execute rights from the target path, not from which binary is running —
identical in spirit to Unix DAC.
### Problem to solve
Today, allowing `/bin/ls` on `~/workspace` also allows `/bin/ls` on `~/... | open | null | false | 7 | [
"enhancement"
] | [] | 2026-03-08T15:15:27Z | 2026-05-04T09:04:00Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | subrih | 3,961,556 | MDQ6VXNlcjM5NjE1NTY= | User | false |
openclaw/openclaw | 4,041,522,900 | I_kwDOQb6kR87w5L7U | 40,009 | https://github.com/openclaw/openclaw/issues/40009 | https://api.github.com/repos/openclaw/openclaw/issues/40009 | [Feature Request]: Replace raw textarea with a structured JSON code editor (with folding support) | ### Summary
### 总结
1. 核心诉求是替换原生 textarea 为支持 JSON 折叠/高亮的代码编辑组件，提升大体积 JSON 内容的编辑体验；
2. 推荐优先使用 CodeMirror v6 或 Monaco Editor 实现，需保证功能适配和样式统一；
3. 验收需覆盖折叠、高亮、格式校验、数据提交等核心场景。
### Problem to solve
# Issue：将 raw 页面的 textarea 替换为支持 JSON 折叠的代码编辑框
<img width="2880" height="1800" alt="Image" src="https://github.com/user-attac... | open | null | false | 2 | [
"enhancement"
] | [] | 2026-03-08T15:34:58Z | 2026-05-04T09:04:02Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | z-ph | 112,166,208 | U_kgDOBq-FQA | User | false |
openclaw/openclaw | 4,041,708,944 | I_kwDOQb6kR87w55WQ | 40,078 | https://github.com/openclaw/openclaw/issues/40078 | https://api.github.com/repos/openclaw/openclaw/issues/40078 | [Feature]: Silent preflight transcription in requireMention channels | ### Summary
Add a per-channel `preflightTranscribe` option that silently transcribes incoming audio messages even when requireMention is enabled, so transcripts are available in history context when the bot is later @mentioned.
### Problem to solve
In requireMention channels, voice messages sent without an @mention ... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-03-08T16:58:47Z | 2026-05-04T08:01:36Z | null | NONE | null | 20260504T090105Z | 2026-05-04T09:01:05Z | p2o51 | 71,637,111 | MDQ6VXNlcjcxNjM3MTEx | User | false |
openclaw/openclaw | 4,329,814,246 | I_kwDOQb6kR88AAAABAhO45g | 71,892 | https://github.com/openclaw/openclaw/issues/71892 | https://api.github.com/repos/openclaw/openclaw/issues/71892 | Feature Request: Auto-switch to vision model when image attachment is detected | # Feature Request: Auto-switch to vision model when image attachment is detected
## Summary
When a user drags/drops or pastes an image into the chat, automatically switch to a configured vision-capable model instead of silently dropping the image or requiring manual intervention.
## Motivation
Currently (per issue ... | closed | not_planned | true | 2 | [] | [] | 2026-04-26T02:13:25Z | 2026-04-29T04:37:25Z | 2026-04-26T10:42:25Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | Guo-bird | 278,681,671 | U_kgDOEJxYRw | User | false |
openclaw/openclaw | 4,329,662,019 | I_kwDOQb6kR88AAAABAhFmQw | 71,849 | https://github.com/openclaw/openclaw/issues/71849 | https://api.github.com/repos/openclaw/openclaw/issues/71849 | Realtime voice consult is too slow/fragile for live calls; add fast memory context path | ## Summary
`voice-call` realtime conversations currently expose `openclaw_agent_consult` directly to the realtime provider via `realtime.toolPolicy`. That is the right conceptual affordance, but in live phone calls it is too slow and brittle for remembered-context questions.
In testing with Twilio + Google Gemini Liv... | closed | completed | true | 2 | [
"enhancement",
"channel: voice-call"
] | [] | 2026-04-26T00:47:36Z | 2026-05-04T04:37:04Z | 2026-05-01T09:12:31Z | CONTRIBUTOR | null | 20260504T060031Z | 2026-05-04T06:00:31Z | amzzzzzzz | 154,392,693 | U_kgDOCTPYdQ | User | false |
openclaw/openclaw | 4,211,588,454 | I_kwDOQb6kR877B71m | 61,858 | https://github.com/openclaw/openclaw/issues/61858 | https://api.github.com/repos/openclaw/openclaw/issues/61858 | [Feature]: Per-agent TTS voice/voiceId configuration | ### Summary
TTS 配置目前是全局的（`messages.tts`），所有 agent 共用同一个音色。在多 agent 架构中，每个 agent 有各自的人设和音色，共享音色会削弱身份辨识度。
### Problem to solve
enhances-personality / multi-agent-voice
### Proposed solution
支持在 agent 配置中覆盖 TTS 参数：
```json
"agents": {
"list": [
{ "id": "main", "tts": { "providers": { "minimax": { "voiceId":... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-04-06T12:21:33Z | 2026-04-28T04:43:00Z | 2026-04-26T02:16:41Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Null-Bug-Null | 104,888,320 | U_kgDOBkB4AA | User | false |
openclaw/openclaw | 4,162,163,837 | I_kwDOQb6kR874FZR9 | 56,701 | https://github.com/openclaw/openclaw/issues/56701 | https://api.github.com/repos/openclaw/openclaw/issues/56701 | Feature request: per-agent TTS voice configuration | ## Description
In a multi-agent setup, all agents share the same TTS voice configured under `messages.tts`. There is no way to assign a different ElevenLabs (or OpenAI) voice per agent.
## Use Case
We run 5 agents in one gateway, each with a distinct persona (different workspace, model, personality). It would be nat... | closed | completed | true | 1 | [] | [] | 2026-03-29T01:36:11Z | 2026-04-28T04:43:00Z | 2026-04-26T02:16:41Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | aceclaw826 | 259,391,320 | U_kgDOD3X_WA | User | false |
openclaw/openclaw | 4,161,622,151 | I_kwDOQb6kR874DVCH | 56,613 | https://github.com/openclaw/openclaw/issues/56613 | https://api.github.com/repos/openclaw/openclaw/issues/56613 | [Feature]: Talk/Voice tab — agent/session switching + per-agent TTS voice | ### Summary
Two related feature requests for the Android (and iOS/macOS) Talk/Voice mode:
## 1. Voice/Talk tab agent switching
Currently the Voice tab is hardcoded to the `main` session (default agent). The Chat tab already has a session picker that lets users switch between agents, but the Voice tab does not.
**Re... | open | null | false | 2 | [] | [] | 2026-03-28T20:49:23Z | 2026-05-01T21:59:55Z | null | NONE | null | 20260502T000037Z | 2026-05-02T00:00:37Z | kenchen3000 | 9,312,409 | MDQ6VXNlcjkzMTI0MDk= | User | false |
openclaw/openclaw | 4,329,723,266 | I_kwDOQb6kR88AAAABAhJVgg | 71,868 | https://github.com/openclaw/openclaw/issues/71868 | https://api.github.com/repos/openclaw/openclaw/issues/71868 | [Bug] Stale images re-hydrated from `[media attached: ...]` text persisting in message history | # [Bug] Stale images re-hydrated from `[media attached: ...]` text persisting in message history
## Summary
`detectAndLoadPromptImages()` re-loads image files by parsing `[media attached: <path>]` markers out of `effectivePrompt`. These markers are injected into **message body text** at ingest time and persist verbat... | closed | completed | true | 4 | [
"bug",
"bug:behavior"
] | [] | 2026-04-26T01:18:49Z | 2026-04-28T04:42:57Z | 2026-04-26T02:18:47Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | jmeadlock | 3,129,201 | MDQ6VXNlcjMxMjkyMDE= | User | false |
openclaw/openclaw | 4,161,810,335 | I_kwDOQb6kR874EC-f | 56,638 | https://github.com/openclaw/openclaw/issues/56638 | https://api.github.com/repos/openclaw/openclaw/issues/56638 | Agent tool surface missing media/audio delivery tool since 2026.3.x | ## Summary
Since upgrading to 2026.3.x (currently on 2026.3.24), agents no longer have a tool to deliver audio files as voice notes to Telegram. The Telegram channel adapter confirms media is supported (openclaw channels capabilities --channel telegram shows media in support flags), but the agent's tool surface has no... | closed | completed | true | 1 | [] | [] | 2026-03-28T22:06:48Z | 2026-04-28T04:42:55Z | 2026-04-26T02:24:56Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | idouglas | 5,369,577 | MDQ6VXNlcjUzNjk1Nzc= | User | false |
openclaw/openclaw | 3,858,919,448 | I_kwDOQb6kR87mAnAY | 2,597 | https://github.com/openclaw/openclaw/issues/2597 | https://api.github.com/repos/openclaw/openclaw/issues/2597 | Context/state lost after unexpected compaction or session reset | # Feature Request: Add context usage percentage to Runtime line
## Summary
Agents have no visibility into their context window usage, causing unexpected compaction and lost state. Adding `context=X%` to the Runtime line would enable proactive state management.
## Problem
When using Clawdbot as a persistent AI assis... | open | null | false | 8 | [
"enhancement"
] | [] | 2026-01-27T05:37:47Z | 2026-05-04T23:09:26Z | null | NONE | null | 20260505T000034Z | 2026-05-05T00:00:34Z | Alvin-MN | 186,787,477 | U_kgDOCyImlQ | User | false |
openclaw/openclaw | 3,882,951,829 | I_kwDOQb6kR87ncSSV | 6,599 | https://github.com/openclaw/openclaw/issues/6599 | https://api.github.com/repos/openclaw/openclaw/issues/6599 | Feature: Add /models test-fallback command to verify fallback chain | ## Summary
Add a command to test that the model fallback chain is correctly configured and working, without waiting for a real failure.
## Motivation
Currently, there's no way to verify fallback configuration works until an actual provider failure occurs. This makes it hard to:
1. Confirm fallback models are correc... | open | null | false | 4 | [
"enhancement"
] | [] | 2026-02-01T21:12:19Z | 2026-05-05T07:15:41Z | null | NONE | null | 20260505T090045Z | 2026-05-05T09:00:45Z | jonwiley | 4,779,831 | MDQ6VXNlcjQ3Nzk4MzE= | User | false |
openclaw/openclaw | 3,883,062,506 | I_kwDOQb6kR87nctTq | 6,625 | https://github.com/openclaw/openclaw/issues/6625 | https://api.github.com/repos/openclaw/openclaw/issues/6625 | Feature: Graceful sub-agent timeout (pre-timeout warning) | ## Problem
When a sub-agent hits `runTimeoutSeconds`, it's killed immediately with no chance to save progress. **All unsaved work is lost** — code, research, analysis, generated content, anything.
## Proposed Solution
Inject a system message to the sub-agent N seconds before the timeout expires:
> ⚠️ You have 60 seco... | open | null | false | 2 | [
"enhancement"
] | [] | 2026-02-01T21:56:06Z | 2026-05-05T00:07:54Z | null | NONE | null | 20260505T030055Z | 2026-05-05T03:00:55Z | jekard | 42,802,518 | MDQ6VXNlcjQyODAyNTE4 | User | false |
openclaw/openclaw | 3,883,500,917 | I_kwDOQb6kR87neYV1 | 6,757 | https://github.com/openclaw/openclaw/issues/6757 | https://api.github.com/repos/openclaw/openclaw/issues/6757 | Feature Request: Agent-triggered context compaction (self-compact tool) | ## Summary
I am Wyatt, an OpenClaw agent autonomously filing this feature request after researching the problem with my human operator.
We would like a way for agents to trigger context compaction on their own session without requiring user intervention.
## Current Behavior
- `/compact` command requires an **author... | open | null | false | 3 | [
"enhancement"
] | [] | 2026-02-02T01:30:52Z | 2026-05-05T03:09:27Z | null | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | jfaithedu | 119,699,344 | U_kgDOByJ3kA | User | false |
openclaw/openclaw | 3,883,581,292 | I_kwDOQb6kR87ner9s | 6,792 | https://github.com/openclaw/openclaw/issues/6792 | https://api.github.com/repos/openclaw/openclaw/issues/6792 | Feature: configPatch in plugin manifest — automatic config setup on install | ## Problem
Installing an OpenClaw plugin that requires config changes is a multi-step, error-prone process:
```bash
# Step 1: Install the package
npm install -g claude-search-proxy
# Step 2: Register the plugin
openclaw plugins install --link $(npm root -g)/claude-search-proxy
# Step 3: Manually edit openclaw.json ... | open | null | false | 3 | [
"enhancement"
] | [] | 2026-02-02T02:17:15Z | 2026-05-05T00:07:57Z | null | NONE | null | 20260505T030055Z | 2026-05-05T03:00:55Z | LePetitPince | 258,336,360 | U_kgDOD2XmaA | User | false |
openclaw/openclaw | 3,883,672,373 | I_kwDOQb6kR87nfCM1 | 6,820 | https://github.com/openclaw/openclaw/issues/6820 | https://api.github.com/repos/openclaw/openclaw/issues/6820 | Add openai-codex/gpt-5.2 to xhigh thinking whitelist | ### Request
Please add `openai-codex/gpt-5.2` to the xhigh thinking whitelist (`XHIGH_MODEL_REFS`).
### Context
In `dist/auto-reply/thinking.js`, xhigh is currently restricted to a small hard-coded set (e.g. `openai/gpt-5.2`, `openai-codex/gpt-5.2-codex`, `openai-codex/gpt-5.1-codex`).
However, many setups (including... | open | null | false | 5 | [
"enhancement"
] | [] | 2026-02-02T02:59:20Z | 2026-05-05T02:55:19Z | null | NONE | null | 20260505T030055Z | 2026-05-05T03:00:55Z | alfadb | 3,182,249 | MDQ6VXNlcjMxODIyNDk= | User | false |
openclaw/openclaw | 3,884,017,744 | I_kwDOQb6kR87ngWhQ | 6,890 | https://github.com/openclaw/openclaw/issues/6890 | https://api.github.com/repos/openclaw/openclaw/issues/6890 | [Feature]: Add Ralph Loop feature and add max iteration number into the agent configuration | ## Summary
Ralph Loop (iteratively feeding the same prompt to an agent until it explicitly chooses STOP) is already implemented in Kimi, but control is limited to global config and a CLI flag (`--max-ralph-iterations`). This request is to let an agent specification define a default Ralph iteration limit so each agent ... | open | null | false | 3 | [
"enhancement"
] | [] | 2026-02-02T05:08:36Z | 2026-05-05T07:57:40Z | null | NONE | null | 20260505T090045Z | 2026-05-05T09:00:45Z | tunglinwood | 113,751,333 | U_kgDOBse1JQ | User | false |
openclaw/openclaw | 3,884,310,224 | I_kwDOQb6kR87nhd7Q | 6,946 | https://github.com/openclaw/openclaw/issues/6946 | https://api.github.com/repos/openclaw/openclaw/issues/6946 | Feature: Processing indicator message for Telegram (send ⌛️ while thinking, delete on reply) | ## Summary
Add a "processing indicator" feature that sends a visible message (e.g., ⌛️) when a message is received and being processed, then deletes it when the reply is sent.
## Motivation
Current feedback mechanisms:
- **Typing indicators** ("typing...") - subtle, easy to miss
- **Ack reactions** - adds emoji to t... | open | null | false | 4 | [
"enhancement"
] | [] | 2026-02-02T06:57:41Z | 2026-05-05T10:31:01Z | null | NONE | null | 20260505T120033Z | 2026-05-05T12:00:33Z | howardpen9 | 36,764,495 | MDQ6VXNlcjM2NzY0NDk1 | User | false |
openclaw/openclaw | 4,199,998,369 | I_kwDOQb6kR876VuOh | 60,236 | https://github.com/openclaw/openclaw/issues/60236 | https://api.github.com/repos/openclaw/openclaw/issues/60236 | [Bug]: Discord voice joins successfully but never captures user speech on OpenClaw 2026.4.2 | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
On OpenClaw 2026.4.2, Discord voice joins successfully and slash commands work, but the bot never detects or captures user speech in the voice channel, while Telegram voice works normally in the same runtime.
... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-03T10:55:37Z | 2026-04-28T04:42:54Z | 2026-04-26T02:27:43Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | phuc-peter-nguyen-covergo | 23,151,994 | MDQ6VXNlcjIzMTUxOTk0 | User | false |
openclaw/openclaw | 3,884,421,446 | I_kwDOQb6kR87nh5FG | 6,966 | https://github.com/openclaw/openclaw/issues/6966 | https://api.github.com/repos/openclaw/openclaw/issues/6966 | Dynamic Model Switching Based on API Rate Limits | ## Problem
Currently, moltbot has a static default model configuration. When Claude API rate limits are hit, the system rotates through configured auth profiles but doesn't automatically switch to a cheaper/faster model to reduce load and token consumption.
**Gap**: No mechanism to auto-downgrade model (opus → sonnet... | open | null | false | 3 | [
"enhancement"
] | [] | 2026-02-02T07:30:37Z | 2026-05-05T00:07:58Z | null | NONE | null | 20260505T030055Z | 2026-05-05T03:00:55Z | dreamingbinary | 56,523 | MDQ6VXNlcjU2NTIz | User | false |
openclaw/openclaw | 3,884,461,774 | I_kwDOQb6kR87niC7O | 6,975 | https://github.com/openclaw/openclaw/issues/6975 | https://api.github.com/repos/openclaw/openclaw/issues/6975 | [Feature] Configurable reply context length for Telegram | ## Summary
When using `[[reply_to_current]]` or `[[reply_to:<id>]]`, the quoted message excerpt in Telegram replies is truncated. A configurable `replyContextLength` parameter would allow users to control how much of the original message is included in replies.
## Use Case
In longer conversations or when replying to d... | closed | completed | true | 6 | [
"enhancement"
] | [] | 2026-02-02T07:42:11Z | 2026-04-30T04:38:14Z | 2026-04-26T06:28:46Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | rex05ai | 258,758,601 | U_kgDOD2xXyQ | User | false |
openclaw/openclaw | 3,884,737,979 | I_kwDOQb6kR87njGW7 | 7,006 | https://github.com/openclaw/openclaw/issues/7006 | https://api.github.com/repos/openclaw/openclaw/issues/7006 | [Feature]: Expose AI model used and usage cost when using openrouter/auto | Problem:
When openrouter/auto is used, the specific model it resolves to is not exposed to the agent, hindering accurate reporting to the user.
Requested Change:
Please ensure the actual model name resolved by openrouter/auto and its cost is accessible to the agent. This would enable reporting in the format [Resolved ... | open | null | false | 3 | [
"enhancement"
] | [] | 2026-02-02T08:50:59Z | 2026-05-05T11:45:40Z | null | NONE | null | 20260505T120033Z | 2026-05-05T12:00:33Z | huzz4561 | 214,878,848 | U_kgDODM7KgA | User | false |
openclaw/openclaw | 3,885,162,303 | I_kwDOQb6kR87nkt8_ | 7,057 | https://github.com/openclaw/openclaw/issues/7057 | https://api.github.com/repos/openclaw/openclaw/issues/7057 | Flaky tests on Windows/WSL: timeouts and ENOENT in pi-tools workspace-paths & safe-bins | ## Summary ✅
When running the test suite under WSL (Ubuntu) I reproduced CI-like results: the build succeeds and the large majority of tests pass, but a small set of tests fail on WSL (likely environment-sensitive / flaky). This issue captures reproduction steps, logs, root-cause hypotheses and suggested mitigations.
... | open | null | false | 3 | [
"enhancement"
] | [] | 2026-02-02T10:27:08Z | 2026-05-05T05:54:17Z | null | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | ThinkIbrokeIt | 122,466,113 | U_kgDOB0yvQQ | User | false |
openclaw/openclaw | 3,886,659,300 | I_kwDOQb6kR87nqbbk | 7,234 | https://github.com/openclaw/openclaw/issues/7234 | https://api.github.com/repos/openclaw/openclaw/issues/7234 | Feature: Granular Discord action gates (split actions.messages into read/edit/delete) | ## Problem
The `channels.discord.actions.messages` config toggle gates `read`, `edit`, `delete`, and `send` together as a single boolean. Setting it to `false` disables all four actions — there's no way to selectively disable just `delete` (or just `edit`, etc.) while keeping the others.
## Use Case
I want to preven... | open | null | false | 2 | [
"enhancement"
] | [] | 2026-02-02T15:48:10Z | 2026-05-05T10:07:42Z | null | NONE | null | 20260505T120033Z | 2026-05-05T12:00:33Z | ryanmcgarvey | 99,396 | MDQ6VXNlcjk5Mzk2 | User | false |
openclaw/openclaw | 3,886,946,711 | I_kwDOQb6kR87nrhmX | 7,281 | https://github.com/openclaw/openclaw/issues/7281 | https://api.github.com/repos/openclaw/openclaw/issues/7281 | Feature: fallbackChannel for hook mappings when primary delivery fails | ## Summary
When a hook mapping has `deliver: true` and the primary channel is unavailable (e.g., WhatsApp disconnected), there's no way to specify an alternative delivery channel. The delivery silently fails and "(error)" is appended to the hook name in session summaries.
## Proposed Solution
Add a `fallbackChannel`... | open | null | false | 2 | [
"enhancement"
] | [] | 2026-02-02T16:45:53Z | 2026-05-05T09:20:49Z | null | NONE | null | 20260505T120033Z | 2026-05-05T12:00:33Z | gmuniz21 | 66,831,691 | MDQ6VXNlcjY2ODMxNjkx | User | false |
openclaw/openclaw | 3,887,327,347 | I_kwDOQb6kR87ns-hz | 7,338 | https://github.com/openclaw/openclaw/issues/7338 | https://api.github.com/repos/openclaw/openclaw/issues/7338 | [Feature]: Agent Attestation Headers | ## Feature Request: Agent Attestation for External API Requests
### Problem Statement
When building services designed specifically for OpenClaw agents (e.g., agent-only APIs, skill backends, agent marketplaces), there is currently no reliable way to verify that an incoming HTTP request originates from an OpenClaw age... | open | null | false | 2 | [
"enhancement"
] | [] | 2026-02-02T18:21:23Z | 2026-05-04T23:09:44Z | null | NONE | null | 20260505T000034Z | 2026-05-05T00:00:34Z | Grenghis-Khan | 63,885,013 | MDQ6VXNlcjYzODg1MDEz | User | false |
openclaw/openclaw | 3,887,421,496 | I_kwDOQb6kR87ntVg4 | 7,359 | https://github.com/openclaw/openclaw/issues/7359 | https://api.github.com/repos/openclaw/openclaw/issues/7359 | [Slack] Agent lacks visibility into own channel messages from DM sessions | ## Problem
The agent can send messages to Slack channels but has no record of those messages when operating in a DM session context. This creates a "blind spot" where the agent:
1. Sends messages to channels (e.g., #ai-agents)
2. Has no memory of sending them when asked in DMs
3. Cannot reference or follow up on its o... | open | null | false | 3 | [
"enhancement"
] | [] | 2026-02-02T18:48:12Z | 2026-05-05T03:09:29Z | null | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | trevorallred | 340,379 | MDQ6VXNlcjM0MDM3OQ== | User | false |
openclaw/openclaw | 3,887,528,163 | I_kwDOQb6kR87ntvjj | 7,379 | https://github.com/openclaw/openclaw/issues/7379 | https://api.github.com/repos/openclaw/openclaw/issues/7379 | Feature Request: Add premium request % to /usage footer | ## Summary
Add GitHub Copilot premium request percentage to the `/usage` footer, similar to how `/status` already displays it.
## Problem
Currently, `/status` shows premium request usage:
```
📊 Usage: Premium 86% left · Chat 100% left
```
However, the `/usage` footer (enabled via `/usage tokens` or `/usage full`) ... | open | null | false | 3 | [
"enhancement"
] | [] | 2026-02-02T19:21:44Z | 2026-05-04T23:09:47Z | null | NONE | null | 20260505T000034Z | 2026-05-05T00:00:34Z | mauk1us | 13,590,929 | MDQ6VXNlcjEzNTkwOTI5 | User | false |
openclaw/openclaw | 3,887,586,493 | I_kwDOQb6kR87nt9y9 | 7,403 | https://github.com/openclaw/openclaw/issues/7403 | https://api.github.com/repos/openclaw/openclaw/issues/7403 | Feature: Private Mode for demos and content creation | ## Summary
Add a **Private Mode** toggle in Mission Control that suppresses all workspace context, allowing users to use OpenClaw for demos, content creation, and public-facing work without exposing personal data or modifying their workspace.
## Use Case
Users creating content, running demos, or sharing their screen... | open | null | false | 3 | [
"enhancement"
] | [] | 2026-02-02T19:43:04Z | 2026-05-05T03:09:30Z | null | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | jacob-creme | 194,879,716 | U_kgDOC52g5A | User | false |
openclaw/openclaw | 3,887,589,611 | I_kwDOQb6kR87nt-jr | 7,406 | https://github.com/openclaw/openclaw/issues/7406 | https://api.github.com/repos/openclaw/openclaw/issues/7406 | [Feature]: Human-readable Telegram topic names in session dropdown | ## Summary
Session dropdown shows raw keys like `agent:main:telegram:group:-123456789:topic:42` instead of human-readable topic names for Telegram forum topics.
## Expected Behavior
Display as `Telegram : GroupName : TopicName` (with spaces around colons, preserving original casing) instead of raw numeric IDs.
## S... | open | null | false | 3 | [
"enhancement"
] | [] | 2026-02-02T19:44:14Z | 2026-05-05T00:08:01Z | null | CONTRIBUTOR | null | 20260505T030055Z | 2026-05-05T03:00:55Z | Diaspar4u | 3,605,840 | MDQ6VXNlcjM2MDU4NDA= | User | false |
openclaw/openclaw | 3,887,662,596 | I_kwDOQb6kR87nuQYE | 7,424 | https://github.com/openclaw/openclaw/issues/7424 | https://api.github.com/repos/openclaw/openclaw/issues/7424 | [Feature]: per Agent Model allow list | ## Summary
I don't want the family to set the family agent's model to opus and kill my openrouter credits
## Proposed solution
Have per agent model allow list
## Alternatives considered
Running multiple gateways but then I have to duplicate everything including skills | open | null | false | 2 | [
"enhancement"
] | [] | 2026-02-02T20:12:19Z | 2026-05-05T09:29:04Z | null | NONE | null | 20260505T120033Z | 2026-05-05T12:00:33Z | kirel | 9,124 | MDQ6VXNlcjkxMjQ= | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.