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 | 4,163,834,011 | I_kwDOQb6kR874LxCb | 56,958 | https://github.com/openclaw/openclaw/issues/56958 | https://api.github.com/repos/openclaw/openclaw/issues/56958 | [Bug]: Model fallback does not actually execute after rate_limit when session is pinned to primary model | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
When the primary model hits `429 rate_limit_error`, OpenClaw correctly decides to use the configured fallback model, but the active session’s persisted/live model selection switches the run back to the original... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-03-29T11:11:00Z | 2026-04-27T05:08:05Z | 2026-04-24T17:56:39Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Xai17 | 64,147,636 | MDQ6VXNlcjY0MTQ3NjM2 | User | false |
openclaw/openclaw | 3,891,004,300 | I_kwDOQb6kR87n7AOM | 7,985 | https://github.com/openclaw/openclaw/issues/7985 | https://api.github.com/repos/openclaw/openclaw/issues/7985 | Feature: Support bypass-AI mode for custom commands to reduce token consumption | ## Problem
Currently, all custom commands defined in or must go through the AI model:
This consumes input/output tokens even for simple, deterministic commands like:
- - just runs a local script and returns output
- - displays system information
- Custom info queries that don't need AI processing
### Example... | closed | completed | true | 4 | [
"enhancement"
] | [] | 2026-02-03T12:21:33Z | 2026-04-27T05:08:04Z | 2026-04-24T17:56:55Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | ShenWang96 | 25,247,671 | MDQ6VXNlcjI1MjQ3Njcx | User | false |
openclaw/openclaw | 3,891,061,744 | I_kwDOQb6kR87n7OPw | 7,993 | https://github.com/openclaw/openclaw/issues/7993 | https://api.github.com/repos/openclaw/openclaw/issues/7993 | [Feature]: Show gateway bundled skills in openclaw skills list | ## Summary
The Control UI shows bundled gateway skills (e.g. `1password`) but `openclaw skills list` returns empty because it only scans managed/workspace skills. This mismatch is confusing and makes the CLI look broken even though the gateway clearly has skills.
## Proposed solution
When the gateway is reachable, i... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-03T12:35:31Z | 2026-04-27T05:08:04Z | 2026-04-24T17:56:59Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | PaulRBerg | 8,782,666 | MDQ6VXNlcjg3ODI2NjY= | User | false |
openclaw/openclaw | 3,891,096,566 | I_kwDOQb6kR87n7Wv2 | 8,000 | https://github.com/openclaw/openclaw/issues/8000 | https://api.github.com/repos/openclaw/openclaw/issues/8000 | Agora channel plugin for relay integration | ## Summary
Integrate Agora relay as a channel in OpenClaw, so agents can participate in Agora conversations with their full cognitive capability (not just dumb auto-responders).
## Current State
- Agora relay works: `wss://agora-relay.lbsa71.net`
- UI works for humans: `npx @rookdaemon/agora-ui`
- But agents need manu... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-02-03T12:44:02Z | 2026-04-27T05:08:03Z | 2026-04-24T17:57:04Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | rookdaemon | 258,400,181 | U_kgDOD2bftQ | User | false |
openclaw/openclaw | 3,891,396,992 | I_kwDOQb6kR87n8gGA | 8,033 | https://github.com/openclaw/openclaw/issues/8033 | https://api.github.com/repos/openclaw/openclaw/issues/8033 | Feature Request: Telegram Inline Keyboard Support | ## Use Case
I need to send reminder messages with interactive buttons to users. For example:
Reminder: Check Discord support reply
Discord account frozen - follow-up needed
• Contacted support via X account
• Waiting for response
[✅ Resolved] [⏰ Remind Later]
When users click buttons, I can automatically handle co... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-03T13:54:04Z | 2026-04-27T05:08:02Z | 2026-04-24T17:57:08Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | rssprivacy-commits | 258,811,864 | U_kgDOD20n2A | User | false |
openclaw/openclaw | 3,891,528,014 | I_kwDOQb6kR87n9AFO | 8,049 | https://github.com/openclaw/openclaw/issues/8049 | https://api.github.com/repos/openclaw/openclaw/issues/8049 | Feature: Add exec hooks to skill metadata for automatic command preprocessing | ## Problem
When using skills that require specific environment setup (e.g., AWS Bedrock integration for Claude Code), the AI agent often "forgets" to apply prerequisite commands even when documented in SKILL.md.
### Our Use Case
We use Claude Code CLI with AWS Bedrock authentication. The `.bashrc` contains required e... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-03T14:22:37Z | 2026-04-27T05:08:02Z | 2026-04-24T17:57:13Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | niceysam | 256,747,835 | U_kgDOD02pOw | User | false |
openclaw/openclaw | 3,891,763,123 | I_kwDOQb6kR87n95ez | 8,079 | https://github.com/openclaw/openclaw/issues/8079 | https://api.github.com/repos/openclaw/openclaw/issues/8079 | [Feature Request] Add proxy configuration option for browser tool | ## Feature Request
### Description
Add a configuration option to specify a proxy server for the browser tool. Currently, when OpenClaw starts Chrome/Chromium, there's no way to configure it to use a proxy server.
### Use Case
- Users behind corporate firewalls need proxy access
- Users in regions with restricted inte... | closed | completed | true | 6 | [
"enhancement"
] | [] | 2026-02-03T15:12:38Z | 2026-04-27T05:08:01Z | 2026-04-24T17:57:17Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | 13967186047lee-maker | 226,995,395 | U_kgDODYesww | User | false |
openclaw/openclaw | 3,891,857,645 | I_kwDOQb6kR87n-Qjt | 8,089 | https://github.com/openclaw/openclaw/issues/8089 | https://api.github.com/repos/openclaw/openclaw/issues/8089 | [Feature]: cache for ci/cd | ## Summary
Caching some deps for ci/cd in github action would be helpful for minimize running time of ci.
## Proposed solution
Considering actions/cache or `cache` parameter of setup-node, etc. | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-03T15:33:52Z | 2026-04-27T05:08:00Z | 2026-04-24T17:57:21Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | hanqyu | 32,977,015 | MDQ6VXNlcjMyOTc3MDE1 | User | false |
openclaw/openclaw | 3,891,989,872 | I_kwDOQb6kR87n-w1w | 8,106 | https://github.com/openclaw/openclaw/issues/8106 | https://api.github.com/repos/openclaw/openclaw/issues/8106 | 🍛 [Feature]: Add Swiggy MCP Skill for Food Ordering (India) | **Summary**
Swiggy, India's leading food delivery platform, recently released official MCP server integrations for their services. This feature proposes adding a new skill that enables OpenClaw to manage Swiggy orders via the existing mcporter CLI.
**Background**
Swiggy launched MCP integrations across their platforms... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-02-03T16:03:46Z | 2026-04-27T05:07:58Z | 2026-04-24T17:57:26Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | AdamParker19 | 42,702,233 | MDQ6VXNlcjQyNzAyMjMz | User | false |
openclaw/openclaw | 3,892,450,754 | I_kwDOQb6kR87oAhXC | 8,185 | https://github.com/openclaw/openclaw/issues/8185 | https://api.github.com/repos/openclaw/openclaw/issues/8185 | [Feature]: Memory flush on /new and /reset (pre-reset memory save) | ## Summary
When a user triggers `/new` or `/reset`, the session is discarded without giving the agent a chance to save important context to memory files. The existing pre-compaction memory flush only triggers on auto-compaction, not on manual resets.
## Problem
- User says `/new` mid-conversation → important context i... | closed | completed | true | 5 | [
"enhancement"
] | [] | 2026-02-03T17:50:13Z | 2026-04-27T04:30:12Z | 2026-04-24T17:57:30Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | NullSense | 10,047,484 | MDQ6VXNlcjEwMDQ3NDg0 | User | false |
openclaw/openclaw | 3,892,502,739 | I_kwDOQb6kR87oAuDT | 8,192 | https://github.com/openclaw/openclaw/issues/8192 | https://api.github.com/repos/openclaw/openclaw/issues/8192 | [Feature]: Workspace Files Editor in Control UI | ## Summary
A web-based editor for workspace files (MEMORY.md, HEARTBEAT.md, SOUL.md, etc.) accessible via the Control UI.
## Problem
Currently, users can edit `openclaw.json` via the Control UI, but there's no way to edit workspace files (agent memory, identity, instructions) from a browser. This forces users to SSH/a... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-02-03T18:04:21Z | 2026-04-27T05:07:57Z | 2026-04-24T17:57:35Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Globarti | 1,726,746 | MDQ6VXNlcjE3MjY3NDY= | User | false |
openclaw/openclaw | 4,324,703,292 | I_kwDOQb6kR88AAAABAcW8PA | 71,190 | https://github.com/openclaw/openclaw/issues/71190 | https://api.github.com/repos/openclaw/openclaw/issues/71190 | [Bug]: mergeOrphanedTrailingUserPrompt snowballs queued messages after abort — model re-answers on every turn | ## Bug type
Bug (incorrect behavior without a crash)
## Summary
`mergeOrphanedTrailingUserPrompt()` snowballs queued messages into an ever-growing single text block after an abort, causing the model to re-answer already-addressed messages on every subsequent turn.
## OpenClaw Version
2026.4.22
## Environment
- m... | closed | completed | true | 2 | [] | [] | 2026-04-24T18:00:04Z | 2026-04-30T04:36:40Z | 2026-04-27T05:25:18Z | CONTRIBUTOR | null | 20260430T060018Z | 2026-04-30T06:00:18Z | aidiffuser | 77,209,370 | MDQ6VXNlcjc3MjA5Mzcw | User | false |
openclaw/openclaw | 4,324,705,997 | I_kwDOQb6kR88AAAABAcXGzQ | 71,191 | https://github.com/openclaw/openclaw/issues/71191 | https://api.github.com/repos/openclaw/openclaw/issues/71191 | [Bug]: openclaw memory index still fails with 'Unknown memory embedding provider: local' on 2026.4.23 after #70836 fix | ## Summary
`openclaw memory index --verbose` still fails with `Unknown memory embedding provider: local` on OpenClaw `2026.4.23`, even though issue #70836 was closed by commit `e069d03945ef384f2be379f00d21cbf6d647de80` (`fix(memory-core): declare local memoryEmbeddingProviders contract (#70873)`).
This looks like one ... | closed | duplicate | true | 1 | [] | [] | 2026-04-24T18:00:32Z | 2026-04-27T05:07:47Z | 2026-04-24T18:14:11Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | mogglemoss | 10,966,195 | MDQ6VXNlcjEwOTY2MTk1 | User | false |
openclaw/openclaw | 3,892,715,871 | I_kwDOQb6kR87oBiFf | 8,215 | https://github.com/openclaw/openclaw/issues/8215 | https://api.github.com/repos/openclaw/openclaw/issues/8215 | Feature: Emit separate "thinking" stream in agent events for real-time reasoning visibility | ## Summary
Currently, reasoning/thinking content is handled via the `onReasoningStream` callback but is not broadcast as an agent event. This makes it impossible for UIs (like the Control UI console view) to display real-time reasoning streams.
## Proposed Change
Add a `"thinking"` stream type to agent events by emi... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-03T18:59:11Z | 2026-04-27T05:07:56Z | 2026-04-24T18:03:55Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | AhmadSid110 | 187,219,946 | U_kgDOCyi_6g | User | false |
openclaw/openclaw | 3,892,823,395 | I_kwDOQb6kR87oB8Vj | 8,231 | https://github.com/openclaw/openclaw/issues/8231 | https://api.github.com/repos/openclaw/openclaw/issues/8231 | [Feature]: terminal-based web search | **Summary**
OpenCalw would be much more useful if it did not require a Brave API key for web search functionality.
**Proposed Solution**
Provide a terminal-based web search skill that works without requiring an external API key.
**Alternatives Considered**
Supporting additional search APIs from other providers (e.... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-03T19:28:12Z | 2026-04-27T05:07:55Z | 2026-04-24T18:03:58Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | R-STEFFES | 230,040,686 | U_kgDODbYkbg | User | false |
openclaw/openclaw | 3,893,007,386 | I_kwDOQb6kR87oCpQa | 8,255 | https://github.com/openclaw/openclaw/issues/8255 | https://api.github.com/repos/openclaw/openclaw/issues/8255 | [Feature]: post-write hooks for workspace files | ## Summary
When the agent writes files to the workspace (memory logs, notes, etc.), there's no built-in way to trigger external actions like:
• Auto-commit to git
• Encrypt sensitive files before push
• Sync to remote storage
• Run custom scripts
## Proposed solution
Add optional hook configuration, something like:... | closed | not_planned | true | 6 | [
"enhancement"
] | [] | 2026-02-03T20:11:50Z | 2026-04-27T05:07:54Z | 2026-04-24T18:04:02Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | 0xOLOR1N | 153,139,705 | U_kgDOCSC5-Q | User | false |
openclaw/openclaw | 3,893,055,191 | I_kwDOQb6kR87oC07X | 8,263 | https://github.com/openclaw/openclaw/issues/8263 | https://api.github.com/repos/openclaw/openclaw/issues/8263 | Feature Request: Webchat User Display Name Customization | **Summary**
Currently, the OpenClaw webchat interface shows "you" as the label for user messages, while the assistant can be customized via `ui.assistant.name`. Users would benefit from being able to customize their display name as well.
**Feature Request**
Add the ability to customize the user display name in the web... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-03T20:24:40Z | 2026-04-27T05:07:54Z | 2026-04-24T18:04:05Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | B-B-Eight | 106,223,076 | U_kgDOBlTV5A | User | false |
openclaw/openclaw | 3,893,332,142 | I_kwDOQb6kR87oD4iu | 8,308 | https://github.com/openclaw/openclaw/issues/8308 | https://api.github.com/repos/openclaw/openclaw/issues/8308 | feat: auto-compaction support for spawned subagent sessions | ## Feature Request
**Current behavior:** Subagent sessions created via `sessions_spawn` run until they hit the context window limit and then stop. They have no access to the compaction safeguard that the main session uses.
**Desired behavior:** Subagent sessions should optionally support auto-compaction, allowing lon... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-03T21:31:49Z | 2026-04-27T05:07:53Z | 2026-04-24T18:04:12Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | 1kuna | 47,035,140 | MDQ6VXNlcjQ3MDM1MTQw | User | false |
openclaw/openclaw | 3,893,750,650 | I_kwDOQb6kR87oFet6 | 8,370 | https://github.com/openclaw/openclaw/issues/8370 | https://api.github.com/repos/openclaw/openclaw/issues/8370 | [Feature Request] Option to echo transcript to user before agent reply | ## Summary
Add a configuration option to send the audio transcript back to the user before the agent processes and replies. This would help users verify what was transcribed, especially for voice-heavy workflows.
## Use Case
When sending voice messages via Telegram (or other channels), I want to see the transcriptio... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-03T23:36:17Z | 2026-04-27T05:07:52Z | 2026-04-24T18:04:16Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | chenzhony | 21,352,097 | MDQ6VXNlcjIxMzUyMDk3 | User | false |
openclaw/openclaw | 3,894,275,964 | I_kwDOQb6kR87oHe98 | 8,485 | https://github.com/openclaw/openclaw/issues/8485 | https://api.github.com/repos/openclaw/openclaw/issues/8485 | Feature: Token count and cost display in ⚡ indicator | ## Problem
Currently the ⚡ lightning bolt indicator shows when a message is being processed, but users have no visibility into token consumption or costs during the conversation.
## Proposed Solution
Enhance the ⚡ indicator to optionally display token count and cumulative cost information:
**Examples:**
- `⚡[1800] ... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-02-04T02:56:53Z | 2026-04-27T05:07:52Z | 2026-04-24T18:04:19Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | dobensonCO | 130,842,624 | U_kgDOB8yAAA | User | false |
openclaw/openclaw | 3,895,678,640 | I_kwDOQb6kR87oM1aw | 8,722 | https://github.com/openclaw/openclaw/issues/8722 | https://api.github.com/repos/openclaw/openclaw/issues/8722 | [Feature Request] Feishu plugin: Add typing indicator via message reactions | ### Problem Statement
The official `@openclaw/feishu` plugin has two UX issues affecting user experience:
#### Issue 1: Missing Typing Indicator
The plugin currently sets `capabilities.reactions = false`, which prevents the typing indicator feature from working. Users receive no visual feedback while AI is processing ... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-02-04T09:58:42Z | 2026-04-27T05:07:51Z | 2026-04-24T18:04:24Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | qi-o | 113,908,052 | U_kgDOBsoZVA | User | false |
openclaw/openclaw | 3,895,852,680 | I_kwDOQb6kR87oNf6I | 8,754 | https://github.com/openclaw/openclaw/issues/8754 | https://api.github.com/repos/openclaw/openclaw/issues/8754 | [Feature]: Agent has no idea of what the gateway dashboard is | ## Summary
Noticed that my agent does not know much about OpenClaw features
It would be helpful to have the OpenClaw docs in agent's context.
## Proposed solution
What would you like Clawdbot to do?
## Alternatives considered
Any other approaches you have considered?
## Additional context
Links, screenshots, or ... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-04T10:39:15Z | 2026-04-27T05:07:50Z | 2026-04-24T18:04:27Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Kalimero31 | 113,331,848 | U_kgDOBsFOiA | User | false |
openclaw/openclaw | 4,324,726,696 | I_kwDOQb6kR88AAAABAcYXqA | 71,192 | https://github.com/openclaw/openclaw/issues/71192 | https://api.github.com/repos/openclaw/openclaw/issues/71192 | Non-default agent models.json caches can stay stale after config model updates | Title: Non-default agent models.json caches can stay stale after config model updates
## Summary
When `models.providers.*` is updated in `openclaw.json`, generated `models.json` for non-default agent directories can remain stale even after restart/reload, while the source config is already correct.
In our case this ... | closed | not_planned | true | 1 | [] | [] | 2026-04-24T18:04:59Z | 2026-04-29T04:41:40Z | 2026-04-26T07:28:05Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | FtlC-ian | 40,432,656 | MDQ6VXNlcjQwNDMyNjU2 | User | false |
openclaw/openclaw | 4,324,817,492 | I_kwDOQb6kR88AAAABAcd6VA | 71,200 | https://github.com/openclaw/openclaw/issues/71200 | https://api.github.com/repos/openclaw/openclaw/issues/71200 | Google Veo video generation succeeds, but OpenClaw fails with 404 when downloading MLDev result URI | ### Summary
Google Veo video generation appears to succeed with a valid Gemini API key, but OpenClaw ultimately fails with `404 Not Found` when handling the generated result.
After tracing the flow, the issue seems to be in the Google video provider adapter: for Gemini MLDev responses, OpenClaw treats the returned `g... | closed | completed | true | 2 | [] | [] | 2026-04-24T18:24:29Z | 2026-04-28T05:26:07Z | 2026-04-25T10:07:42Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | panhaishan | 40,199,485 | MDQ6VXNlcjQwMTk5NDg1 | User | false |
openclaw/openclaw | 3,895,866,171 | I_kwDOQb6kR87oNjM7 | 8,758 | https://github.com/openclaw/openclaw/issues/8758 | https://api.github.com/repos/openclaw/openclaw/issues/8758 | Feature request: Send stickers in Telegram | ## Description
It would be great if the bot could send Telegram stickers via `sendSticker` API.
##... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-04T10:42:16Z | 2026-04-27T05:07:43Z | 2026-04-24T18:31:58Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | karinakomyak-tech | 249,317,814 | U_kgDODtxJtg | User | false |
openclaw/openclaw | 3,895,887,221 | I_kwDOQb6kR87oNoV1 | 8,761 | https://github.com/openclaw/openclaw/issues/8761 | https://api.github.com/repos/openclaw/openclaw/issues/8761 | Feature: Allow MEDIA: directive with absolute paths (opt-in) | ## Feature Request
Allow `MEDIA:` directives to use absolute paths or arbitrary paths, with an opt-in configuration to bypass the current relative-path-only restriction.
## Current Behavior
The `isValidMedia()` function in `src/media/parse.ts` only accepts:
- HTTP/HTTPS URLs
- Relative paths starting with `./` (no `... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-04T10:46:43Z | 2026-04-27T05:07:42Z | 2026-04-24T18:32:03Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | phuctm97 | 25,026,967 | MDQ6VXNlcjI1MDI2OTY3 | User | false |
openclaw/openclaw | 3,896,046,229 | I_kwDOQb6kR87oOPKV | 8,781 | https://github.com/openclaw/openclaw/issues/8781 | https://api.github.com/repos/openclaw/openclaw/issues/8781 | Atlas support: use ChatGPT UI for browser-scoped tasks to reduce API usage | ## Summary
Add full ChatGPT Atlas support and route browser-scoped work through Atlas when available to reduce API usage, with safe fallback to existing browser providers.
## Motivation
Atlas uses ChatGPT in the browser (not API-based models), so routing browser-scoped tasks through Atlas can save API usage while keep... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-02-04T11:24:57Z | 2026-04-27T05:07:41Z | 2026-04-24T18:32:07Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | winning-z | 112,189,745 | U_kgDOBq_hMQ | User | false |
openclaw/openclaw | 3,896,228,815 | I_kwDOQb6kR87oO7vP | 8,801 | https://github.com/openclaw/openclaw/issues/8801 | https://api.github.com/repos/openclaw/openclaw/issues/8801 | feat: Add on_gateway_ready hook to run tasks immediately after gateway starts | ## Problem
After a gateway restart, there's no way to immediately trigger agent actions. Cron jobs only run at their scheduled intervals, which can cause delays of several minutes before the agent responds to pending messages.
## Use Case
- After restart, check for unanswered messages and respond immediately
- Run s... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-04T12:08:28Z | 2026-04-27T05:07:41Z | 2026-04-24T18:32:11Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | invius95 | 178,498,804 | U_kgDOCqOs9A | User | false |
openclaw/openclaw | 3,896,798,877 | I_kwDOQb6kR87oRG6d | 8,865 | https://github.com/openclaw/openclaw/issues/8865 | https://api.github.com/repos/openclaw/openclaw/issues/8865 | Feature Request: TUI Light Theme / Color Customization Support | ## Feature Request
### Problem
When using a light terminal theme (e.g., Catppuccin Latte, Solarized Light), the OpenClaw TUI has very low contrast and is difficult to read. The TUI appears to be designed primarily for dark backgrounds.
### Environment
- Terminal: Blink Shell (iPad)
- Terminal Theme: Catppuccin Latte ... | closed | completed | true | 8 | [
"enhancement",
"tui"
] | [] | 2026-02-04T14:21:47Z | 2026-04-27T05:07:40Z | 2026-04-24T18:32:16Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | lil-woogie-min-ai | 258,885,735 | U_kgDOD25IZw | User | false |
openclaw/openclaw | 3,897,066,045 | I_kwDOQb6kR87oSII9 | 8,886 | https://github.com/openclaw/openclaw/issues/8886 | https://api.github.com/repos/openclaw/openclaw/issues/8886 | [Feature]: WhatsApp documents (PDF) not downloaded - mediaPath is null | *Summary*
WhatsApp inbound documents (PDFs) are not downloaded. Unlike images and voice notes which download correctly, documents arrive with mediaPath: null and only show the <media:document> placeholder.
*Environment*
- OpenClaw version: 2026.1.30
- Node: v22.22.0
- WhatsApp config: mediaMaxMb: 50
*Expected:* Docu... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-04T15:20:25Z | 2026-04-27T05:07:39Z | 2026-04-24T18:32:20Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Krish-Bajaj | 53,073,541 | MDQ6VXNlcjUzMDczNTQx | User | false |
openclaw/openclaw | 3,897,115,063 | I_kwDOQb6kR87oSUG3 | 8,895 | https://github.com/openclaw/openclaw/issues/8895 | https://api.github.com/repos/openclaw/openclaw/issues/8895 | RFC: Batch state primitives for orchestration layer | ## Problem
The orchestration layer treats each agent run as independent, even when they're part of a larger batch operation. This causes problems when:
1. **Compaction occurs mid-batch** — The agent loses track of what's been done
2. **Timeout kills a run** — No partial progress is preserved
3. **Sub-agent coordinati... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-02-04T15:30:27Z | 2026-04-27T05:07:39Z | 2026-04-24T18:32:24Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | joetomasone | 56,984,887 | MDQ6VXNlcjU2OTg0ODg3 | User | false |
openclaw/openclaw | 3,897,146,329 | I_kwDOQb6kR87oSbvZ | 8,901 | https://github.com/openclaw/openclaw/issues/8901 | https://api.github.com/repos/openclaw/openclaw/issues/8901 | Feature: Separate tool event streaming from channel verbose mode | ## Problem
The `verboseDefault` config setting currently controls two distinct behaviors:
1. **Channel verbose output** — Sending tool call details as messages in Telegram/Discord/etc.
2. **WebSocket event broadcasting** — Emitting `agent` events with `stream: "tool"` to connected operator/webchat clients.
When `ver... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-02-04T15:37:21Z | 2026-04-27T05:07:38Z | 2026-04-24T18:32:28Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | a-b13 | 257,067,647 | U_kgDOD1KKfw | User | false |
openclaw/openclaw | 3,897,269,411 | I_kwDOQb6kR87oS5yj | 8,920 | https://github.com/openclaw/openclaw/issues/8920 | https://api.github.com/repos/openclaw/openclaw/issues/8920 | [Feature]: Better UI for configuration | ## Summary
Hi,
Maybe I am the only one who is stupid for using the OpenClaw project but either way when I want to add new agents or if I am on the Config page then hard to add new channels and there are many other things which from the UI nightmare to setup. When I am looking the Config page then all of the settings l... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-02-04T16:05:02Z | 2026-04-27T05:07:37Z | 2026-04-24T18:32:32Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | StrangerGithuber | 1,939,827 | MDQ6VXNlcjE5Mzk4Mjc= | User | false |
openclaw/openclaw | 3,897,468,380 | I_kwDOQb6kR87oTqXc | 8,945 | https://github.com/openclaw/openclaw/issues/8945 | https://api.github.com/repos/openclaw/openclaw/issues/8945 | Feature Request: Per-Channel Model Defaults | ## Problem
Currently, OpenClaw only supports:
- **Global primary model** (`agents.defaults.model.primary`) - affects all sessions
- **Per-session overrides** (via `/model` command) - temporary, resets on `/new`
**What's missing:** Persistent per-channel model defaults.
### Use Case
User wants different models for d... | closed | completed | true | 4 | [
"enhancement"
] | [] | 2026-02-04T16:51:46Z | 2026-04-27T05:07:36Z | 2026-04-24T18:32:37Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | reechard-ai | 258,124,991 | U_kgDOD2Ksvw | User | false |
openclaw/openclaw | 4,324,871,753 | I_kwDOQb6kR88AAAABAchOSQ | 71,211 | https://github.com/openclaw/openclaw/issues/71211 | https://api.github.com/repos/openclaw/openclaw/issues/71211 | Security: exec tool returns raw stdout/stderr to agent without secret redaction | **Bug Report: `exec` tool returns raw stdout/stderr to agent without secret redaction**
### Summary
Foreground `exec` results in OpenClaw are assembled directly from aggregated command output and returned to the agent without an obvious secret-redaction pass. If a command prints secrets, those secrets can enter intern... | open | null | false | 1 | [] | [] | 2026-04-24T18:35:35Z | 2026-05-03T06:07:29Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | nhaener | 16,600,979 | MDQ6VXNlcjE2NjAwOTc5 | User | false |
openclaw/openclaw | 4,324,876,679 | I_kwDOQb6kR88AAAABAchhhw | 71,212 | https://github.com/openclaw/openclaw/issues/71212 | https://api.github.com/repos/openclaw/openclaw/issues/71212 | Slack download-file action fails silently on non-image files (DOCX, PDF, etc.) | ## Bug Report
### Description
The message(action="download-file") tool action fails silently when downloading non-image files (DOCX, PDF, CSV, etc.) from Slack. The tool reports success but the file is not usable by the agent.
### Root Cause
In dist/common-*.js, the imageResultFromFile() function forces all downloade... | closed | completed | true | 1 | [] | [] | 2026-04-24T18:36:43Z | 2026-04-27T04:46:54Z | 2026-04-24T23:56:12Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | teamrazo | 97,799,015 | U_kgDOBdRLZw | User | false |
openclaw/openclaw | 4,324,867,707 | I_kwDOQb6kR88AAAABAcg-ew | 71,208 | https://github.com/openclaw/openclaw/issues/71208 | https://api.github.com/repos/openclaw/openclaw/issues/71208 | [Bug]: config.patch triggers gateway restart when openclaw.json contains ${VAR} env-refs on gateway.* or plugins.* paths | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Every config.patch RPC triggers a full SIGUSR1 gateway restart on instances whose openclaw.json uses ${VAR} env-ref syntax for any gateway.* or plugins.* field. The config change itself only touches an unrelated hot-reloadab... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-04-24T18:34:42Z | 2026-04-27T04:45:37Z | 2026-04-25T00:27:21Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | robbiethompson18 | 216,262,277 | U_kgDODOPmhQ | User | false |
openclaw/openclaw | 3,899,172,207 | I_kwDOQb6kR87oaKVv | 9,239 | https://github.com/openclaw/openclaw/issues/9239 | https://api.github.com/repos/openclaw/openclaw/issues/9239 | [Feature Request] Add option to disable sub-agent tool activity notifications in Telegram | ## Summary
When a sub-agent is spawned via `sessions_spawn`, each tool call (Browser, Web Fetch, Exec, etc.) sends a notification message to the Telegram chat. This creates significant noise when sub-agents perform complex tasks.
## Current Behavior
With `channels.telegram.streamMode: "off"` configured, the main age... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-02-05T01:35:55Z | 2026-04-27T05:07:35Z | 2026-04-24T18:50:40Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | dotfile-adam | 259,463,990 | U_kgDOD3cbNg | User | false |
openclaw/openclaw | 3,899,186,430 | I_kwDOQb6kR87oaNz- | 9,244 | https://github.com/openclaw/openclaw/issues/9244 | https://api.github.com/repos/openclaw/openclaw/issues/9244 | [Feature]: Cost-Optimized LLM Gateway for OpenClaw | # Feature Request: Cost-Optimized LLM Gateway for OpenClaw
## Summary
Implement an intelligent LLM Gateway to optimize API costs and improve response quality through smart routing, multi-tier caching, diff-only responses, and security controls.
Update Feb 11: since there are many pending requests i vibe coded the g... | closed | not_planned | true | 10 | [
"enhancement"
] | [] | 2026-02-05T01:42:55Z | 2026-04-27T05:07:35Z | 2026-04-24T18:50:44Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | sp4cerat | 12,142,588 | MDQ6VXNlcjEyMTQyNTg4 | User | false |
openclaw/openclaw | 3,899,262,494 | I_kwDOQb6kR87oagYe | 9,264 | https://github.com/openclaw/openclaw/issues/9264 | https://api.github.com/repos/openclaw/openclaw/issues/9264 | Feature Request: Cross-Channel Context Sharing | ## Cross-Channel Context Sharing Feature Request
### Problem Description
Currently, OpenClaw maintains separate session contexts for different chat channels (private chat vs group chat), so context from one channel is not available in another. This creates discontinuity when users want to continue conversations across... | closed | not_planned | true | 2 | [
"enhancement"
] | [] | 2026-02-05T02:20:31Z | 2026-04-27T05:07:34Z | 2026-04-24T18:50:47Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | botkey40-stack | 258,830,264 | U_kgDOD21vuA | User | false |
openclaw/openclaw | 3,899,285,203 | I_kwDOQb6kR87oal7T | 9,272 | https://github.com/openclaw/openclaw/issues/9272 | https://api.github.com/repos/openclaw/openclaw/issues/9272 | Feature: Remote session reset command for crash recovery | ## Problem
When a session crashes due to context overflow (e.g., prompt exceeds model limits), the gateway becomes unresponsive for that session. The only recovery path currently requires:
1. SSH/terminal access to the host machine
2. Manually deleting or moving the corrupted session file from `~/.openclaw/sessions/`... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-02-05T02:31:40Z | 2026-04-27T05:07:33Z | 2026-04-24T18:50:51Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | firstmovr | 155,568,852 | U_kgDOCUXK1A | User | false |
openclaw/openclaw | 3,899,405,111 | I_kwDOQb6kR87obDM3 | 9,298 | https://github.com/openclaw/openclaw/issues/9298 | https://api.github.com/repos/openclaw/openclaw/issues/9298 | [Feature]: Support for Jami messenger | ## Summary
Support the Jami open source messenger (https://jami.net/).
## Proposed solution
Ability to integrate Jami messenger: https://docs.jami.net/en_US/developer/new-developers/apis-of-jami.html
## Alternatives considered
None.
## Additional context
N/A
| closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-02-05T03:28:58Z | 2026-04-27T05:07:32Z | 2026-04-24T18:50:54Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | rmoeritz-alayacare | 213,376,875 | U_kgDODLffaw | User | false |
openclaw/openclaw | 3,899,655,895 | I_kwDOQb6kR87ocAbX | 9,372 | https://github.com/openclaw/openclaw/issues/9372 | https://api.github.com/repos/openclaw/openclaw/issues/9372 | [Feature]: Add Cisco Webex Channel Integration | ## Summary
I'd like to propose adding Webex (Cisco Webex) as a new channel integration for Moltbot.
## Motivation
- Webex is widely used in enterprise environments, especially by Cisco customers
- Many teams use Webex for daily communication and would benefit from Moltbot integration
- Currently no Webex support exi... | closed | not_planned | true | 4 | [
"enhancement"
] | [] | 2026-02-05T04:57:22Z | 2026-04-27T05:07:32Z | 2026-04-24T18:50:58Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Khalil12138 | 150,124,128 | U_kgDOCPK2YA | User | false |
openclaw/openclaw | 3,899,717,482 | I_kwDOQb6kR87ocPdq | 9,386 | https://github.com/openclaw/openclaw/issues/9386 | https://api.github.com/repos/openclaw/openclaw/issues/9386 | [Feature]: Working Memory for Long-Horizon Agent Tasks - local user-space, markdown solution | ## Summary
Agents suffer from **catastrophic context loss** after session compaction, making long-horizon tasks unreliable. We propose a simple solution based on hierarchical memory architecture, and a PR to fix it in core. Interim solution uses existing hooks and does not change the core repository (local markdown, ... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-02-05T05:18:10Z | 2026-04-27T05:07:31Z | 2026-04-24T18:51:01Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | onelouder | 13,673,203 | MDQ6VXNlcjEzNjczMjAz | User | false |
openclaw/openclaw | 3,899,817,169 | I_kwDOQb6kR87ocnzR | 9,402 | https://github.com/openclaw/openclaw/issues/9402 | https://api.github.com/repos/openclaw/openclaw/issues/9402 | Feature Request: Pre-message model router hook | ## Summary
Add a hook that allows dynamic model selection **before** each message is processed, enabling automatic routing to cheaper models for simple tasks.
## Problem
Currently, model selection is static (per-agent or per-session). Users pay Opus prices for simple tasks like:
- "What time is it?"
- "ok"
- "thanks... | closed | completed | true | 3 | [
"enhancement"
] | [] | 2026-02-05T05:50:07Z | 2026-04-27T05:07:30Z | 2026-04-24T18:51:05Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | ario | 952,509 | MDQ6VXNlcjk1MjUwOQ== | User | false |
openclaw/openclaw | 3,900,018,675 | I_kwDOQb6kR87odY_z | 9,448 | https://github.com/openclaw/openclaw/issues/9448 | https://api.github.com/repos/openclaw/openclaw/issues/9448 | Feature: Expose chatType in Runtime Context | # Feature Request: Expose `chatType` in Runtime Context
## Summary
The agent runtime currently exposes channel information via the `## Runtime` section of the system prompt, but does not include the `chatType` (direct/group/channel) information. This makes it impossible for agents to conditionally adjust behavior base... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-05T07:01:03Z | 2026-04-27T05:07:29Z | 2026-04-24T18:51:08Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | jxdbaby | 249,119,425 | U_kgDODtlCwQ | User | false |
openclaw/openclaw | 3,901,003,422 | I_kwDOQb6kR87ohJae | 9,567 | https://github.com/openclaw/openclaw/issues/9567 | https://api.github.com/repos/openclaw/openclaw/issues/9567 | RFC: Semantic Compression Protocol (SCP) — A Token-Efficient Lingua Franca for AI Perception | ### RFC: Semantic Compression Protocol (SCP) — A Token-Efficient "Lingua Franca" for AI Perception
### 1. The Problem: Information Entropy vs. Compute Cost
Currently, agents like OpenClaw interact with the world (specifically UIs and code) through highly inefficient mediums:
- **Raw HTML/DOM**: Extremely noisy. To fin... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-02-05T11:08:14Z | 2026-04-27T05:07:28Z | 2026-04-24T18:51:12Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Strelitzia-reginae | 76,481,705 | MDQ6VXNlcjc2NDgxNzA1 | User | false |
openclaw/openclaw | 3,901,538,881 | I_kwDOQb6kR87ojMJB | 9,618 | https://github.com/openclaw/openclaw/issues/9618 | https://api.github.com/repos/openclaw/openclaw/issues/9618 | Feature: per-agent allowlist for cross-agent memory read access | Use case: a "master" agent should be able to read selected other agents’ memories. Memory is currently scoped per-agent, which is fine by default, but there’s no opt-in way to read from another agent.
Proposal:
- Per-agent allowlist config:
agents.<id>.memory.allowReadFrom: ["agentA", "agentB", "*"]
* means all ot... | closed | completed | true | 3 | [
"enhancement"
] | [] | 2026-02-05T13:17:44Z | 2026-04-27T05:07:27Z | 2026-04-24T18:51:15Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Helmi | 175,346 | MDQ6VXNlcjE3NTM0Ng== | User | false |
openclaw/openclaw | 3,902,229,858 | I_kwDOQb6kR87ol01i | 9,696 | https://github.com/openclaw/openclaw/issues/9696 | https://api.github.com/repos/openclaw/openclaw/issues/9696 | Feature Request: Display cost estimates in session_status | # Feature Request: Display cost estimates in session_status
## Summary
Add cost estimation to `/status` and `session_status` output based on local pricing config and LiteLLM database.
## Problem
Currently, `session_status` shows token usage but no cost information:
```
🧮 Tokens: 84k in / 396 out
```
Users need to m... | closed | completed | true | 1 | [
"enhancement",
"dedupe:parent"
] | [] | 2026-02-05T15:42:05Z | 2026-04-27T05:07:26Z | 2026-04-24T18:51:19Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Emcooo | 56,719,572 | MDQ6VXNlcjU2NzE5NTcy | User | false |
openclaw/openclaw | 3,902,469,774 | I_kwDOQb6kR87omvaO | 9,738 | https://github.com/openclaw/openclaw/issues/9738 | https://api.github.com/repos/openclaw/openclaw/issues/9738 | Feature Request: Add Google Cloud Speech-to-Text and Text-to-Speech support | ## Summary
Add support for Google Cloud Speech-to-Text and Text-to-Speech APIs as providers for audio transcription and TTS.
## Motivation
- Users with GCP accounts already have these APIs enabled and want to use their $300 free trial credits
- GCP offers generous free tiers:
- **Speech-to-Text**: 60 minutes/month... | closed | not_planned | true | 2 | [
"enhancement"
] | [] | 2026-02-05T16:30:41Z | 2026-04-27T05:07:26Z | 2026-04-24T18:51:23Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | TeigenZhang | 38,639,187 | MDQ6VXNlcjM4NjM5MTg3 | User | false |
openclaw/openclaw | 3,902,545,054 | I_kwDOQb6kR87onBye | 9,753 | https://github.com/openclaw/openclaw/issues/9753 | https://api.github.com/repos/openclaw/openclaw/issues/9753 | Can agents prove why they act? PIC plugin idea for OpenClaw | Hey OpenClaw folks 👋
Curious if anyone would find this useful: I’ve been hacking on a plugin that forces agents to *show their reasoning* before using powerful tools: shell, write_file, send_email, etc.
It’s based on the **PIC Standard (Provenance & Intent Contracts)**:
Every tool call must include a JSON “contract”... | closed | not_planned | true | 7 | [
"enhancement"
] | [] | 2026-02-05T16:48:08Z | 2026-04-27T05:07:25Z | 2026-04-24T18:51:27Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | madeinplutofabio | 20,666,378 | MDQ6VXNlcjIwNjY2Mzc4 | User | false |
openclaw/openclaw | 3,902,718,739 | I_kwDOQb6kR87onsMT | 9,777 | https://github.com/openclaw/openclaw/issues/9777 | https://api.github.com/repos/openclaw/openclaw/issues/9777 | Add --quiet / -q flag to suppress doctor warnings in CLI output | ## Problem
When running `openclaw sessions list --json`, the output includes doctor warnings before the JSON:
```
│
◇ Doctor warnings ─────────────────────────────────────────╮
│ │
│ - State dir migration skipped: target already exists │
│ (/Users/jon... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-05T17:31:07Z | 2026-04-27T05:07:24Z | 2026-04-24T18:51:30Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | jontsai | 422,501 | MDQ6VXNlcjQyMjUwMQ== | User | false |
openclaw/openclaw | 3,903,059,746 | I_kwDOQb6kR87oo_ci | 9,837 | https://github.com/openclaw/openclaw/issues/9837 | https://api.github.com/repos/openclaw/openclaw/issues/9837 | [Feature]: Support Anthropic adaptive thinking and effort parameter for Opus 4.6 | ## Summary
Anthropic's Claude Opus 4.6 introduces **adaptive thinking** (`thinking.type: "adaptive"`) as the recommended way to use extended thinking. The previous `thinking.type: "enabled"` + `budget_tokens` approach is deprecated on Opus 4.6 and will be removed in a future model release.
OpenClaw currently only sup... | closed | completed | true | 13 | [
"enhancement"
] | [] | 2026-02-05T18:57:25Z | 2026-04-27T05:07:24Z | 2026-04-24T18:51:35Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | shaharsha | 11,346,965 | MDQ6VXNlcjExMzQ2OTY1 | User | false |
openclaw/openclaw | 3,903,364,694 | I_kwDOQb6kR87oqJ5W | 9,871 | https://github.com/openclaw/openclaw/issues/9871 | https://api.github.com/repos/openclaw/openclaw/issues/9871 | [Feature]: Expose sessionKey to PluginCommandContext | ## Summary
Plugin commands registered via `registerCommand()` cannot access the current session because `PluginCommandContext` does not include `sessionKey`. This prevents plugin commands from reading or writing per-session state (e.g., toggling a per-session setting, checking session status, or operating on the curre... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-02-05T20:05:41Z | 2026-04-27T05:07:23Z | 2026-04-24T18:51:38Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | leiyangyou | 104,397 | MDQ6VXNlcjEwNDM5Nw== | User | false |
openclaw/openclaw | 3,903,444,624 | I_kwDOQb6kR87oqdaQ | 9,884 | https://github.com/openclaw/openclaw/issues/9884 | https://api.github.com/repos/openclaw/openclaw/issues/9884 | Feature: Per-agent default accountId for outbound messages | # Feature Request: Per-Agent Default `accountId` for Outbound Messages
## Problem
When an agent has its own Discord account (via `channels.discord.accounts.<accountId>`), the binding correctly routes **inbound** messages to that agent. However, **outbound** messages sent via the `message` tool default to the `default... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-02-05T20:25:49Z | 2026-04-27T05:07:22Z | 2026-04-24T18:51:42Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | electronicshaman | 124,746,599 | U_kgDOB297Zw | User | false |
openclaw/openclaw | 3,903,527,661 | I_kwDOQb6kR87oqxrt | 9,891 | https://github.com/openclaw/openclaw/issues/9891 | https://api.github.com/repos/openclaw/openclaw/issues/9891 | Feature request: cosmetic display for silent replies (NO_REPLY) | DRAFT / NON-FINAL
# Feature request: Cosmetic display for silent replies (NO_REPLY)
## Problem
When the agent returns the silent sentinel token `NO_REPLY`, some UIs/logs display it (often truncated as "NO"), which is confusing/noisy for the operator.
Observed surfaces:
- WebUI transcript/log view shows truncated `NO... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-05T20:48:24Z | 2026-04-27T05:07:21Z | 2026-04-24T18:51:46Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | cthornsburg | 58,783,988 | MDQ6VXNlcjU4NzgzOTg4 | User | false |
openclaw/openclaw | 3,903,751,260 | I_kwDOQb6kR87oroRc | 9,920 | https://github.com/openclaw/openclaw/issues/9920 | https://api.github.com/repos/openclaw/openclaw/issues/9920 | Feature Request: Rules injection for persistent agent behavior | ## Problem
Agents forget critical patterns despite having them documented in workspace files (AGENTS.md, MEMORY.md). Files are long, get skimmed, and patterns get lost in session context.
Even when rules are placed at the top of AGENTS.md, they get ignored because:
- AGENTS.md is ~300 lines — agents skim it
- Session... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-02-05T21:48:53Z | 2026-04-27T05:07:21Z | 2026-04-24T18:51:49Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | jgottapu | 7,075,860 | MDQ6VXNlcjcwNzU4NjA= | User | false |
openclaw/openclaw | 3,904,078,059 | I_kwDOQb6kR87os4Dr | 9,979 | https://github.com/openclaw/openclaw/issues/9979 | https://api.github.com/repos/openclaw/openclaw/issues/9979 | [Feature Request] Add Support for Google Cloud Vertex AI Provider | ### 🚀 Feature Description
I would like to suggest adding **Google Cloud Vertex AI** as a supported LLM provider for OpenClaw. This will allow users to leverage the Google Cloud ecosystem and its latest models with high rate limits and advanced configurations.
### 💡 Why this is needed
Adding a Vertex AI provider will... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-05T23:40:05Z | 2026-04-27T05:07:20Z | 2026-04-24T18:51:52Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | smturtle2 | 74,640,768 | MDQ6VXNlcjc0NjQwNzY4 | User | false |
openclaw/openclaw | 3,904,085,984 | I_kwDOQb6kR87os5_g | 9,982 | https://github.com/openclaw/openclaw/issues/9982 | https://api.github.com/repos/openclaw/openclaw/issues/9982 | Feature Request: Option to disable Discord status bar refresh button | ## Description
The Discord status bar (showing context %, sub-agents, etc.) includes a refresh/reload button (🔄 icon) that can cause issues when clicked during config restarts or active operations.
## Request
Add a config option to disable the refresh button on the Discord status bar, e.g.:
```json
{
"channels":... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-05T23:43:03Z | 2026-04-27T05:07:19Z | 2026-04-24T18:51:56Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | RookDevNinja | 257,829,444 | U_kgDOD14qRA | User | false |
openclaw/openclaw | 3,904,251,700 | I_kwDOQb6kR87otic0 | 10,010 | https://github.com/openclaw/openclaw/issues/10010 | https://api.github.com/repos/openclaw/openclaw/issues/10010 | Feature Request: Agent Teams - Parallel Agent Coordination | ## Summary
Inspired by Claude Code's experimental Agent Teams feature, I'd like to propose a similar capability for OpenClaw that enables parallel agent coordination with inter-agent communication.
## Current State
OpenClaw already has building blocks:
- `sessions_spawn`: Creates independent agent sessions
- `sessio... | closed | not_planned | true | 14 | [
"enhancement"
] | [
"BradGroux"
] | 2026-02-06T00:54:30Z | 2026-04-27T05:07:18Z | 2026-04-24T18:52:00Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | watari-ai | 258,264,784 | U_kgDOD2TO0A | User | false |
openclaw/openclaw | 3,904,377,261 | I_kwDOQb6kR87ouBGt | 10,032 | https://github.com/openclaw/openclaw/issues/10032 | https://api.github.com/repos/openclaw/openclaw/issues/10032 | Feature: Multi-user profile system with automatic profile building from conversations | ## Summary
Request for a built-in system to create and maintain multiple user profiles that automatically grow stronger as conversations progress.
## Use Cases
### Group Chats
- When multiple people interact in a group chat (e.g., iMessage, Telegram groups), the agent should automatically:
- Identify unique users ... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-02-06T01:42:59Z | 2026-04-27T05:07:18Z | 2026-04-24T18:52:03Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | realsamrat | 59,194,708 | MDQ6VXNlcjU5MTk0NzA4 | User | false |
openclaw/openclaw | 3,904,477,760 | I_kwDOQb6kR87ouZpA | 10,050 | https://github.com/openclaw/openclaw/issues/10050 | https://api.github.com/repos/openclaw/openclaw/issues/10050 | [Feature]: Implement "Zero-Log" Secure Secret Handoff to Prevent Telemetry Leakage | ## Summary
As OpenClaw grows in popularity, a wider range of users—including those who are less tech-savvy—are adopting the platform. Many of these users rely on free or third-party model providers that collect chat logs for telemetry and training. Currently, when agents exchange sensitive data (API keys, tokens, crede... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-02-06T02:17:30Z | 2026-04-27T05:07:17Z | 2026-04-24T18:52:07Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | CrimsonDevil333333 | 49,252,920 | MDQ6VXNlcjQ5MjUyOTIw | User | false |
openclaw/openclaw | 3,904,600,758 | I_kwDOQb6kR87ou3q2 | 10,067 | https://github.com/openclaw/openclaw/issues/10067 | https://api.github.com/repos/openclaw/openclaw/issues/10067 | Feature: Cron job catch-up for missed runs after gateway restart | ## Problem
When the gateway restarts after a cron job's scheduled time has passed, the job is skipped and waits for the next cycle. This causes missed runs if the gateway happens to be down during a job's scheduled time.
**Example scenario:**
- Job scheduled for 08:00 daily
- Gateway restarts at 09:35 (e.g., after an... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-06T03:02:49Z | 2026-04-27T05:07:16Z | 2026-04-24T18:52:10Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | TeigenZhang | 38,639,187 | MDQ6VXNlcjM4NjM5MTg3 | User | false |
openclaw/openclaw | 3,904,621,439 | I_kwDOQb6kR87ou8t_ | 10,070 | https://github.com/openclaw/openclaw/issues/10070 | https://api.github.com/repos/openclaw/openclaw/issues/10070 | [Feature]: canvas:a2ui:bundle script is not Windows compatible - should use Node.js instead of bash | ## Summary
The `canvas:a2ui:bundle` script currently uses `bash scripts/bundle-a2ui.sh` command, which is not friendly for Windows development environments and prevents successful builds on Windows. This limits the project's cross-platform development capabilities.
## Proposed solution
Migrate the `canvas:a2ui:bundl... | closed | completed | true | 3 | [
"enhancement"
] | [] | 2026-02-06T03:11:12Z | 2026-04-27T05:07:15Z | 2026-04-24T18:52:14Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | JAVA-LW | 58,633,930 | MDQ6VXNlcjU4NjMzOTMw | User | false |
openclaw/openclaw | 3,904,735,882 | I_kwDOQb6kR87ovYqK | 10,101 | https://github.com/openclaw/openclaw/issues/10101 | https://api.github.com/repos/openclaw/openclaw/issues/10101 | [Feature]: Feishu/Lark: Support reading quoted/replied message content (reply/quote context) so the agent can see the referenced “chat history” | # Feishu/Lark – include quoted/replied message content (“会话记录/引用回复”) in inbound events
## Title
Feishu/Lark: Support reading quoted/replied message content (reply/quote context) so the agent can see the referenced “chat history”
## Summary
In Feishu/Lark chats, users often use **Reply/Quote** (引用回复) or forward a “ch... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-02-06T03:54:00Z | 2026-04-27T05:07:14Z | 2026-04-24T18:52:18Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | lllyys | 221,728,326 | U_kgDODTdORg | User | false |
openclaw/openclaw | 3,904,991,213 | I_kwDOQb6kR87owW_t | 10,141 | https://github.com/openclaw/openclaw/issues/10141 | https://api.github.com/repos/openclaw/openclaw/issues/10141 | Feature: Display current installed version in Web UI Dashboard | **Problem**
The Web UI Dashboard Overview currently displays:
- `wizard.lastRunVersion` - Last version that ran the configuration wizard
- `meta.lastTouchedVersion` - Last version that modified the config file
But it does **not** display the **actual installed OpenClaw version** (from `npm list -g openclaw`).
This c... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-02-06T05:35:05Z | 2026-04-27T05:07:14Z | 2026-04-24T18:52:22Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | anthony-raton | 258,748,344 | U_kgDOD2wvuA | User | false |
openclaw/openclaw | 3,905,044,686 | I_kwDOQb6kR87owkDO | 10,149 | https://github.com/openclaw/openclaw/issues/10149 | https://api.github.com/repos/openclaw/openclaw/issues/10149 | [Feature]: Adopt @anthropic-ai/claude-agent-sdk for enhanced agent capabilities | ## Description
OpenClaw currently uses direct Anthropic API calls for Claude interactions. Anthropic now provides an official **Claude Agent SDK** (`@anthropic-ai/claude-agent-sdk`) that wraps Claude Code CLI and offers advanced agentic capabilities.
## Benefits
1. **In-process MCP servers** - Define custom tools as... | closed | not_planned | true | 5 | [
"enhancement"
] | [] | 2026-02-06T05:50:40Z | 2026-04-27T05:07:13Z | 2026-04-24T18:52:26Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | eciuca | 3,948,642 | MDQ6VXNlcjM5NDg2NDI= | User | false |
openclaw/openclaw | 3,905,109,902 | I_kwDOQb6kR87owz-O | 10,161 | https://github.com/openclaw/openclaw/issues/10161 | https://api.github.com/repos/openclaw/openclaw/issues/10161 | [Feature]: Integrate service-ability-creator for Automated AI Service Generation in Clawdbot | ## Summary
**Problem Statement:**
In the current OpenClaw Agent ecosystem, service discovery and inter-agent collaboration rely heavily on manual documentation review or pre-defined interfaces, resulting in:
- **High Integration Costs:** Each new service requires manual interpretation and custom integration logic
- ... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-02-06T06:09:42Z | 2026-04-27T05:07:12Z | 2026-04-24T18:52:29Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | appcom2016 | 38,750,346 | MDQ6VXNlcjM4NzUwMzQ2 | User | false |
openclaw/openclaw | 3,905,119,385 | I_kwDOQb6kR87ow2SZ | 10,164 | https://github.com/openclaw/openclaw/issues/10164 | https://api.github.com/repos/openclaw/openclaw/issues/10164 | Feature: Native Temporal integration for durable workflows and scheduling | ## Summary
Integrate Temporal.io as a first-class orchestration layer in OpenClaw for durable agent workflows, scheduling, and webhook routing.
## Motivation
Currently, OpenClaw has:
- Built-in cron jobs (simple scheduling)
- Webhook hooks (trigger agents)
- `sessions_spawn` (fire-and-forget agent spawning)
These w... | closed | not_planned | true | 2 | [
"enhancement"
] | [] | 2026-02-06T06:12:12Z | 2026-04-27T05:07:11Z | 2026-04-24T18:52:33Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | spk-alex | 205,000,749 | U_kgDODDgQLQ | User | false |
openclaw/openclaw | 3,905,125,154 | I_kwDOQb6kR87ow3si | 10,166 | https://github.com/openclaw/openclaw/issues/10166 | https://api.github.com/repos/openclaw/openclaw/issues/10166 | Feature: API rate limit handling (HTTP 429 backoff + load balancing) | ## Problem
When using Opus or other high-token models with intensive operations (browser automation, large doc generation), Anthropic's tokens-per-minute rate limit triggers HTTP 429 responses.
## Current Behavior
- Gateway passes 429 errors directly to sessions
- No automatic retry or backoff
- Agent must handle paci... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-06T06:13:46Z | 2026-04-27T05:07:11Z | 2026-04-24T18:52:36Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | openmolthi | 259,130,635 | U_kgDOD3IFCw | User | false |
openclaw/openclaw | 3,905,292,544 | I_kwDOQb6kR87oxgkA | 10,212 | https://github.com/openclaw/openclaw/issues/10212 | https://api.github.com/repos/openclaw/openclaw/issues/10212 | Feature Request: Boot hook or startup script support for durable runtime patches | We need to make the current webchat hook message patch durable across restarts and OpenClaw upgrades. Right now the fix appears to be a hot edit in installed `dist/` code, which will be overwritten.
## Goal
On every gateway start (and after any OpenClaw update), automatically re-apply the patch to the installed OpenC... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-06T07:07:32Z | 2026-04-27T05:07:10Z | 2026-04-24T18:52:39Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | gregmeenahan-bot | 259,800,550 | U_kgDOD3w95g | User | false |
openclaw/openclaw | 3,905,584,793 | I_kwDOQb6kR87oyn6Z | 10,268 | https://github.com/openclaw/openclaw/issues/10268 | https://api.github.com/repos/openclaw/openclaw/issues/10268 | Support Copilot hosted tools (web_search_preview) for free Bing search | ## Summary
GitHub Copilot's API supports web search via the `web_search_options` parameter in Chat Completions requests. OpenClaw doesn't pass this parameter, so Copilot users miss out on free Bing web search that the Copilot CLI and VS Code already have.
## Tested & Confirmed Working
We patched `pi-ai`'s `buildPara... | closed | not_planned | true | 2 | [
"enhancement"
] | [] | 2026-02-06T08:33:06Z | 2026-04-27T05:07:09Z | 2026-04-24T18:52:43Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | ChristianFJung | 4,527,481 | MDQ6VXNlcjQ1Mjc0ODE= | User | false |
openclaw/openclaw | 3,905,809,016 | I_kwDOQb6kR87ozep4 | 10,300 | https://github.com/openclaw/openclaw/issues/10300 | https://api.github.com/repos/openclaw/openclaw/issues/10300 | [Android] Handle ERROR_TOO_MANY_REQUESTS (11) in speech recognizer | ## Summary
Android's SpeechRecognizer can return error code 11 (`ERROR_TOO_MANY_REQUESTS`) when recognition is started/stopped rapidly. Current code shows "Speech error (11)" and retries after 600ms, which can make it worse.
## Environment
- Device: Pixel 8a (Android 14)
- Trigger: Rapid start/stop of voice recognitio... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-02-06T09:31:34Z | 2026-04-27T05:07:08Z | 2026-04-24T18:52:47Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | dianeclawd | 259,826,268 | U_kgDOD3yiXA | User | false |
openclaw/openclaw | 3,906,666,810 | I_kwDOQb6kR87o2wE6 | 10,393 | https://github.com/openclaw/openclaw/issues/10393 | https://api.github.com/repos/openclaw/openclaw/issues/10393 | [Feature Request] memory_search: support local file search without external API | ## Problem
The `memory_search` tool currently requires an external embedding API to function. For agents with local file-based memory, this creates unnecessary external dependencies.
## Proposed Solution
1. **BM25/TF-IDF fallback** - When no embedding provider configured, use keyword-based search
2. **Local embeddin... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-02-06T13:04:33Z | 2026-04-27T05:07:08Z | 2026-04-24T18:52:56Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | chelixbot | 259,865,778 | U_kgDOD308sg | User | false |
openclaw/openclaw | 3,907,047,688 | I_kwDOQb6kR87o4NEI | 10,448 | https://github.com/openclaw/openclaw/issues/10448 | https://api.github.com/repos/openclaw/openclaw/issues/10448 | Feature request: HTTP webhook to trigger agent tasks directly | ## Summary
Allow external services to trigger agent tasks via HTTP webhook, without needing to relay through a messaging channel (Telegram/Discord/etc).
## Problem
Currently, to trigger an agent task from an external dashboard/service:
1. Dashboard creates a task with a job ID
2. Job ID gets sent as a message to Teleg... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-02-06T14:30:38Z | 2026-04-27T05:07:07Z | 2026-04-24T18:53:00Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | chad229 | 257,673,683 | U_kgDOD1vJ0w | User | false |
openclaw/openclaw | 3,907,215,723 | I_kwDOQb6kR87o42Fr | 10,479 | https://github.com/openclaw/openclaw/issues/10479 | https://api.github.com/repos/openclaw/openclaw/issues/10479 | [Feature]: Support compaction.customInstructions for auto-compaction | ## Summary
Auto-compaction currently uses a hardcoded project-management style summary format (Goal/Progress/Done/Blocked/Key Decisions/Next Steps). This works great for coding workflows, but loses important context for **casual conversation and companion use cases**.
## Problem
The compaction prompt is designed for... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-02-06T15:12:56Z | 2026-04-27T05:07:06Z | 2026-04-24T18:53:04Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | p697 | 51,314,472 | MDQ6VXNlcjUxMzE0NDcy | User | false |
openclaw/openclaw | 3,907,334,342 | I_kwDOQb6kR87o5TDG | 10,502 | https://github.com/openclaw/openclaw/issues/10502 | https://api.github.com/repos/openclaw/openclaw/issues/10502 | Feature Request: Add tool:call and tool:result Hook Events | ## Feature Request: Add `tool:call` and `tool:result` Hook Events
### Is your feature request related to a problem?
Yes. Currently, the OpenClaw Hook system only supports `gateway:startup` and `agent:message` events. This limits the ability of hooks to:
1. **Track tool execution lifecycle** - Cannot monitor when too... | closed | not_planned | true | 3 | [
"enhancement"
] | [] | 2026-02-06T15:45:58Z | 2026-04-27T05:07:05Z | 2026-04-24T18:53:07Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | whtoo | 909,131 | MDQ6VXNlcjkwOTEzMQ== | User | false |
openclaw/openclaw | 3,907,484,493 | I_kwDOQb6kR87o53tN | 10,524 | https://github.com/openclaw/openclaw/issues/10524 | https://api.github.com/repos/openclaw/openclaw/issues/10524 | Feature: Post-compaction system event injection for context continuity | ## Problem
When an agent's context window fills up and gets compacted, the agent often loses track of what was being discussed. The instructions telling it how to handle compaction also get compressed, leading to:
- Silent responses (NO_REPLY)
- Confused "what were we working on?" responses
- User having to re-expla... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-02-06T16:25:08Z | 2026-04-27T05:07:05Z | 2026-04-24T18:53:11Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | theghostybot | 258,442,348 | U_kgDOD2eEbA | User | false |
openclaw/openclaw | 3,907,667,472 | I_kwDOQb6kR87o6kYQ | 10,547 | https://github.com/openclaw/openclaw/issues/10547 | https://api.github.com/repos/openclaw/openclaw/issues/10547 | [Proposal] Anchor Memory Protocol (AMP): Intelligent Session Refactoring for Pseudo-Infinite Context | ### **Title: [Proposal/RFC] Anchor Memory Protocol (AMP): Intelligent Context Management through Async Session Refactoring & Semantic Indexing**
### **Executive Summary**
We propose the **Anchor Memory Protocol (AMP)**, a novel architecture to solve the "Context Decay" problem in long-running AI agent sessions. By dec... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-06T17:08:21Z | 2026-04-27T05:07:04Z | 2026-04-24T18:53:14Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Strelitzia-reginae | 76,481,705 | MDQ6VXNlcjc2NDgxNzA1 | User | false |
openclaw/openclaw | 3,907,712,243 | I_kwDOQb6kR87o6vTz | 10,553 | https://github.com/openclaw/openclaw/issues/10553 | https://api.github.com/repos/openclaw/openclaw/issues/10553 | [Feature Request] Add MemOS Memory Integration | ## Summary
Request native MemOS (MemoryOS) integration for long-term memory storage and retrieval in OpenClaw.
## Motivation
Current OpenClaw memory system relies on session-based context. MemOS provides:
- Persistent long-term memory across sessions
- User preference tracking
- Factual memory storage
- Memory safet... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-02-06T17:20:33Z | 2026-04-27T05:07:03Z | 2026-04-24T18:53:18Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | veithly | 6,553,213 | MDQ6VXNlcjY1NTMyMTM= | User | false |
openclaw/openclaw | 4,324,955,006 | I_kwDOQb6kR88AAAABAcmTfg | 71,216 | https://github.com/openclaw/openclaw/issues/71216 | https://api.github.com/repos/openclaw/openclaw/issues/71216 | Config schema: add `sandbox`, `routing.rules`, `instances`, and `gateway.nodes.denyPaths` | # Proposed upstream issue — openclaw/openclaw
**Title:** Config schema: add `sandbox`, `routing.rules`, `instances`, and `gateway.nodes.denyPaths`
---
## Context
Running OpenClaw v2026.4.22 on macOS (Mac Studio M4 Max) as a single-user local gateway at `127.0.0.1:18789`. Using it for a mixed cloud / local-Ollama se... | open | null | false | 1 | [] | [] | 2026-04-24T18:53:20Z | 2026-05-03T06:07:30Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | newbrevity | 269,703,465 | U_kgDOEBNZKQ | User | false |
openclaw/openclaw | 3,907,766,401 | I_kwDOQb6kR87o68iB | 10,566 | https://github.com/openclaw/openclaw/issues/10566 | https://api.github.com/repos/openclaw/openclaw/issues/10566 | Feature request: proportional humanDelay mode (based on message length) | ## Summary
Add a `proportional` mode to `humanDelay` that calculates delay based on the length of the outgoing message, simulating more realistic typing speed.
## Current behavior
`humanDelay` supports:
- `off` - no delay
- `natural` - random 800-2500ms
- `custom` - fixed min/max range
All modes use a fixed delay r... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-02-06T17:35:39Z | 2026-04-27T05:07:03Z | 2026-04-24T18:53:21Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | mbilba | 99,972,784 | U_kgDOBfV2sA | User | false |
openclaw/openclaw | 3,908,689,274 | I_kwDOQb6kR87o-d16 | 10,740 | https://github.com/openclaw/openclaw/issues/10740 | https://api.github.com/repos/openclaw/openclaw/issues/10740 | Feature Request: Auto-compact at context threshold | ## Summary
Automatically trigger context compaction when usage exceeds a configurable threshold (e.g., 50%).
## Problem
Currently, agents can only:
1. Check context usage via session_status
2. Alert the user
3. Request user to run `/compact`
This leads to situations where context hits 80-90%+ before anyone notices, r... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-02-06T22:35:52Z | 2026-04-27T05:07:02Z | 2026-04-24T18:53:25Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | steve-glaser-ops | 258,502,852 | U_kgDOD2hwxA | User | false |
openclaw/openclaw | 3,908,706,001 | I_kwDOQb6kR87o-h7R | 10,741 | https://github.com/openclaw/openclaw/issues/10741 | https://api.github.com/repos/openclaw/openclaw/issues/10741 | [Feature Request] Session resume prompt after gateway reconnect | ### Problem
After a gateway disconnect/reconnect (e.g., service restart, network hiccup), the agent "goes silent" even if there was work in progress. The user has to manually prompt to continue.
### Current behavior
1. Gateway disconnects during a task
2. Gateway reconnects
3. Agent waits silently
4. User must ask "... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-02-06T22:40:46Z | 2026-04-27T05:07:01Z | 2026-04-24T18:53:29Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | AIWerk | 258,244,526 | U_kgDOD2R_rg | User | false |
openclaw/openclaw | 3,908,776,641 | I_kwDOQb6kR87o-zLB | 10,747 | https://github.com/openclaw/openclaw/issues/10747 | https://api.github.com/repos/openclaw/openclaw/issues/10747 | Feature: CLI Agent Support for sessions_spawn | ## Summary
Add support for external CLI tools (like OpenAI Codex CLI) as spawnable agents via `sessions_spawn`.
## Current Behavior
`sessions_spawn` only works with OpenClaw's internal agents. External CLI tools like Codex must be run via `exec` and manually monitored.
## Proposed Behavior
Allow defining CLI-based... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-02-06T23:08:26Z | 2026-04-27T05:07:00Z | 2026-04-24T18:53:32Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | jules-agent | 259,457,211 | U_kgDOD3cAuw | User | false |
openclaw/openclaw | 3,908,919,480 | I_kwDOQb6kR87o_WC4 | 10,780 | https://github.com/openclaw/openclaw/issues/10780 | https://api.github.com/repos/openclaw/openclaw/issues/10780 | add browser view via link when GUI on vps not available | I think it would be cool if we could see what openclaw does on the browser live even when it's on a vps which usually doesn't have a desktop and I think that could be useful | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-07T00:19:51Z | 2026-04-27T05:06:59Z | 2026-04-24T18:53:35Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | scorpion7slayer | 120,050,493 | U_kgDOByfTPQ | User | false |
openclaw/openclaw | 3,909,332,429 | I_kwDOQb6kR87pA63N | 10,880 | https://github.com/openclaw/openclaw/issues/10880 | https://api.github.com/repos/openclaw/openclaw/issues/10880 | [Feature]: Support Intel (x86_64) Macs for the macOS desktop app | ## Summary
The OpenClaw macOS desktop app appears to be Apple Silicon–only at the moment. I’d like to request official support for Intel (x86_64) Macs so users on older hardware can run the desktop companion app.
## Proposed solution
- Provide a universal macOS build (arm64 + x86_64), or
- Publish a separate Intel (x8... | closed | completed | true | 3 | [
"enhancement"
] | [] | 2026-02-07T04:11:21Z | 2026-04-27T05:06:59Z | 2026-04-24T18:53:39Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Mokoby | 68,408,564 | MDQ6VXNlcjY4NDA4NTY0 | User | false |
openclaw/openclaw | 3,909,509,197 | I_kwDOQb6kR87pBmBN | 10,929 | https://github.com/openclaw/openclaw/issues/10929 | https://api.github.com/repos/openclaw/openclaw/issues/10929 | [Feature]: WhatsApp voice notes (OGG/Opus) fail transcription via Whisper, needs automatic transcoding to WAV | ## Summary
When receiving WhatsApp voice notes via the gateway, audio is saved correctly but Whisper transcription fails. From debugging gateway logs and media inspection, WhatsApp voice messages arrive as `audio/ogg; codecs=opus`. These files are passed downstream directly to Whisper, which appears to fail decoding (... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-07T06:01:12Z | 2026-04-27T05:06:58Z | 2026-04-24T18:53:43Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | all4one-max | 60,205,465 | MDQ6VXNlcjYwMjA1NDY1 | User | false |
openclaw/openclaw | 3,909,530,008 | I_kwDOQb6kR87pBrGY | 10,936 | https://github.com/openclaw/openclaw/issues/10936 | https://api.github.com/repos/openclaw/openclaw/issues/10936 | Feature Request: Add usage stats to CLI JSON output | ## Use case
We're building automation that calls `openclaw agent --json` programmatically and need to track token usage for monitoring and cost optimization.
## Current behavior
```bash
openclaw agent --message "..." --json
```
Returns:
```json
{
"reply": "Agent response text"
}
```
Usage stats are only available ... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-07T06:14:27Z | 2026-04-27T05:06:56Z | 2026-04-24T18:53:46Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Ratzzz33 | 47,256,936 | MDQ6VXNlcjQ3MjU2OTM2 | User | false |
openclaw/openclaw | 3,909,627,870 | I_kwDOQb6kR87pCC_e | 10,954 | https://github.com/openclaw/openclaw/issues/10954 | https://api.github.com/repos/openclaw/openclaw/issues/10954 | Feature Request: Add WhatsApp message history length limit configuration | ## Feature Request: Add WhatsApp message history length limit configuration
### Problem
Currently, OpenClaw loads all WhatsApp message history into the conversation context, which causes:
1. Context overflow when conversations become too long
2. Users have to manually delete WhatsApp conversation history to reduce co... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-07T06:59:49Z | 2026-04-27T05:06:55Z | 2026-04-24T18:53:50Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | dongfangxiehe-boop | 257,130,146 | U_kgDOD1N-og | User | false |
openclaw/openclaw | 3,909,647,217 | I_kwDOQb6kR87pCHtx | 10,959 | https://github.com/openclaw/openclaw/issues/10959 | https://api.github.com/repos/openclaw/openclaw/issues/10959 | [Feature]: LinkedIn Integration – Expected Timeline? |
### Body
Hi OpenClaw team 👋
First off, great work on OpenClaw — really enjoying experimenting with it so far.
I wanted to ask if there’s any planned timeline or roadmap for LinkedIn integration (for example, sending DMs or interacting with LinkedIn accounts via OpenClaw). This would be a really useful feature f... | closed | not_planned | true | 3 | [
"enhancement"
] | [] | 2026-02-07T07:07:18Z | 2026-04-27T05:06:55Z | 2026-04-24T18:53:53Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | mthirumalai2905 | 98,790,479 | U_kgDOBeNsTw | User | false |
openclaw/openclaw | 3,909,668,636 | I_kwDOQb6kR87pCM8c | 10,964 | https://github.com/openclaw/openclaw/issues/10964 | https://api.github.com/repos/openclaw/openclaw/issues/10964 | Docs: Add model-level usage summary via CodexBar CLI | ### What
Add a beginner-friendly guide and script for summarizing per-model usage
(Codex / Claude) using CodexBar's local cost logs.
This includes:
- current (most recent) model usage
- full per-model breakdown
- text and JSON output options
### Why
OpenClaw users often want to understand:
- which model is being us... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-02-07T07:17:29Z | 2026-04-27T05:06:54Z | 2026-04-24T18:53:57Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | rohansachinpatil | 172,933,149 | U_kgDOCk7AHQ | User | false |
openclaw/openclaw | 3,909,837,516 | I_kwDOQb6kR87pC2LM | 11,025 | https://github.com/openclaw/openclaw/issues/11025 | https://api.github.com/repos/openclaw/openclaw/issues/11025 | Auto-compress images before API calls to prevent 5MB limit errors | ## Problem
When using screenshot/browser tools, images can exceed Anthropic's 5MB limit, causing errors like:
```
LLM request rejected: messages.377.content.0.tool_result.content.1.image.source.base64:
image exceeds 5 MB maximum: 5786592 bytes > 5242880 bytes
```
This happens especially on high-DPI/Retina displays ... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-02-07T09:03:07Z | 2026-04-27T05:06:53Z | 2026-04-24T18:54:00Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | maxschlecht2000-code | 234,594,624 | U_kgDODfuhQA | User | false |
openclaw/openclaw | 3,910,212,496 | I_kwDOQb6kR87pERuQ | 11,133 | https://github.com/openclaw/openclaw/issues/11133 | https://api.github.com/repos/openclaw/openclaw/issues/11133 | Feature request: WebChat/Canvas image upload UI + unified attachment flow across channels | ## Context
OpenClaw already has solid attachment foundations:
- The gateway protocol's `chat.send` schema includes an `attachments` field (`src/gateway/protocol/schema/logs-chat.ts`)
- `parseMessageWithAttachments` (`src/gateway/chat-attachments.ts`) validates base64 content, sniffs MIME types, enforces size limits (... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-07T12:52:58Z | 2026-04-27T05:06:52Z | 2026-04-24T18:54:04Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | antoniarco | 2,615,296 | MDQ6VXNlcjI2MTUyOTY= | User | false |
openclaw/openclaw | 3,910,296,954 | I_kwDOQb6kR87pEmV6 | 11,156 | https://github.com/openclaw/openclaw/issues/11156 | https://api.github.com/repos/openclaw/openclaw/issues/11156 | [Docs]: Remove references to "clawrouter" — routing is built into openclaw gateway | ## Summary
`clawrouter` is referenced in various places (docs, community discussions), but does not exist as a standalone binary. All routing/proxy functionality is built into `openclaw gateway` via the BlockRun plugin.
## Evidence
```bash
$ command -v clawrouter
# → not found (exit 1)
$ which openclaw
/opt/homebre... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-02-07T13:37:18Z | 2026-04-27T05:06:52Z | 2026-04-24T18:54:08Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | BotPuppetMaster | 259,342,082 | U_kgDOD3U_Ag | User | false |
openclaw/openclaw | 3,910,439,852 | I_kwDOQb6kR87pFJOs | 11,197 | https://github.com/openclaw/openclaw/issues/11197 | https://api.github.com/repos/openclaw/openclaw/issues/11197 | Feature: notify pairing requests via Telegram channel | ## Summary
Trying to find a streamline workflow to how I can get pairing request to my telegram instead of CLI or WebUI
## Proposed solution
Add gateway.devices.notifyChannel = telegram config option
## Alternatives considered
Cron job- Check pending.json every minute, ping telegram if new request come.
| closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-07T14:45:14Z | 2026-04-27T05:06:51Z | 2026-04-24T18:54:12Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | pparth230 | 95,541,303 | U_kgDOBbHYNw | User | false |
openclaw/openclaw | 3,910,455,039 | I_kwDOQb6kR87pFM7_ | 11,202 | https://github.com/openclaw/openclaw/issues/11202 | https://api.github.com/repos/openclaw/openclaw/issues/11202 | [Security] Model catalog with resolved apiKey values injected into LLM prompt context on every turn | ## Summary
The runtime model catalog (resolved from `openclaw.json` providers) is serialized into the LLM request payload as part of the system prompt context on every turn. Because environment variable references (`${VAR}`) are resolved to plaintext **before** serialization, all provider API keys are sent to whicheve... | closed | completed | true | 6 | [
"bug",
"security"
] | [] | 2026-02-07T14:51:17Z | 2026-04-27T05:06:50Z | 2026-04-24T18:54:16Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | rplakas | 1,306,838 | MDQ6VXNlcjEzMDY4Mzg= | User | false |
openclaw/openclaw | 3,910,693,739 | I_kwDOQb6kR87pGHNr | 11,264 | https://github.com/openclaw/openclaw/issues/11264 | https://api.github.com/repos/openclaw/openclaw/issues/11264 | [Feature]: add support for XMTP protocol and Convos App | ## Summary
This is a feature request to add support for XMTP protocol and Convos messaging app as supported channels. The PR is here: https://github.com/openclaw/openclaw/pull/10506
## Proposed solution
Allow openclaw to be configured to work on XMTP messaging protocol and Convos providing MLS encrypted messaging.
... | closed | not_planned | true | 3 | [
"enhancement"
] | [] | 2026-02-07T16:42:20Z | 2026-04-27T05:06:49Z | 2026-04-24T18:54:19Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | tantodefi | 107,724,252 | U_kgDOBmu93A | User | false |
openclaw/openclaw | 3,910,892,455 | I_kwDOQb6kR87pG3un | 11,335 | https://github.com/openclaw/openclaw/issues/11335 | https://api.github.com/repos/openclaw/openclaw/issues/11335 | Camera snap: allow targeting specific camera by device ID | ## Feature Request
**Current behavior:**
`camera_snap` only supports `front`/`back` facing selectors. On macOS, all external cameras report as `position: unspecified`, so there's no way to target a specific camera.
**`camera_list` already returns device IDs:**
```json
[
{"name": "MacBook Pro Camera", "id": "6C70704... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-07T18:20:07Z | 2026-04-27T05:06:49Z | 2026-04-24T18:54:23Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | dp-pcs | 203,848,779 | U_kgDODCZ8Sw | User | false |
openclaw/openclaw | 3,910,996,537 | I_kwDOQb6kR87pHRI5 | 11,376 | https://github.com/openclaw/openclaw/issues/11376 | https://api.github.com/repos/openclaw/openclaw/issues/11376 | Feature: Slack placeholder message with post-then-edit pattern | ## Summary
Add a "thinking indicator" for Slack that posts a placeholder message immediately, then edits it with the final response.
## Implementation Pattern (from OpenClaw)
### 1. On message receive — post placeholder immediately:
```javascript
const result = await slack.chat.postMessage({
channel: channelId,
... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-02-07T19:15:37Z | 2026-04-27T05:06:48Z | 2026-04-24T18:54:26Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Airfresh-joey | 234,839,703 | U_kgDODf9elw | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.