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 | 4,066,360,705 | I_kwDOQb6kR87yX72B | 44,294 | https://github.com/openclaw/openclaw/issues/44294 | https://api.github.com/repos/openclaw/openclaw/issues/44294 | Preserve structured ACP backend error kinds instead of mapping all errors to `end_turn` | ## Problem
In `src/acp/translator.ts`, ACP events with `state === "error"` are currently finished as `end_turn` because the protocol lacks a structured error kind.
That means timeouts, rate limits, backend crashes, and deliberate refusals are all effectively collapsed into the same client-visible stop reason.
## Why... | open | null | false | 2 | [] | [] | 2026-03-12T18:07:00Z | 2026-05-04T12:46:50Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | aniruddhaadak80 | 127,435,065 | U_kgDOB5iBOQ | User | false |
openclaw/openclaw | 4,066,360,712 | I_kwDOQb6kR87yX72I | 44,296 | https://github.com/openclaw/openclaw/issues/44296 | https://api.github.com/repos/openclaw/openclaw/issues/44296 | Show the actual PowerShell profile path during onboarding shell-completion setup | ## Problem
`resolveProfileHint()` in `src/wizard/onboarding.completion.ts` falls back to `$PROFILE` for PowerShell instead of resolving the actual profile path.
## Why It Matters
- New Windows users get a generic hint instead of a concrete file path.
- The reload guidance is less actionable than the zsh/bash/fish br... | open | null | false | 2 | [] | [] | 2026-03-12T18:07:00Z | 2026-05-04T13:18:35Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | aniruddhaadak80 | 127,435,065 | U_kgDOB5iBOQ | User | false |
openclaw/openclaw | 4,066,360,714 | I_kwDOQb6kR87yX72K | 44,297 | https://github.com/openclaw/openclaw/issues/44297 | https://api.github.com/repos/openclaw/openclaw/issues/44297 | Surface Slack external arg-menu fallback as a visible health signal | ## Problem
In `src/slack/monitor/slash.ts`, failures during external arg-menu registration are treated as best-effort and silently downgraded to static menus unless verbose logs are enabled.
## Why It Matters
- Provider startup can partially degrade without a visible warning.
- Operators may not realize external arg... | open | null | false | 2 | [] | [] | 2026-03-12T18:07:00Z | 2026-05-04T12:46:53Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | aniruddhaadak80 | 127,435,065 | U_kgDOB5iBOQ | User | false |
openclaw/openclaw | 4,066,455,702 | I_kwDOQb6kR87yYTCW | 44,309 | https://github.com/openclaw/openclaw/issues/44309 | https://api.github.com/repos/openclaw/openclaw/issues/44309 | [Feature]: Add one-way dispatch mode for A2A handoffs without reply-back ping-pong | ## Summary
Add a dispatch-only / handoff mode for agent-to-agent message delivery that drops a task into the target session without reply-back ping-pong or announce back to the requester.
## Problem to solve
Current `sessions_send` semantics are great for agent-to-agent conversation, but they are a poor fit for a ve... | open | null | false | 3 | [] | [] | 2026-03-12T18:25:29Z | 2026-05-04T12:46:56Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | Kaze310 | 183,121,490 | U_kgDOCuo2Ug | User | false |
openclaw/openclaw | 4,066,802,599 | I_kwDOQb6kR87yZnun | 44,347 | https://github.com/openclaw/openclaw/issues/44347 | https://api.github.com/repos/openclaw/openclaw/issues/44347 | Google Chat: Add threaded reply support + receive all messages in spaces | ## Summary
Two feature requests for the Google Chat channel adapter to bring it closer to parity with other channels (Telegram, Discord, Slack).
---
### 1. Threaded Replies
**Current behavior:** When a user sends a message in a Google Chat space, the inbound metadata correctly includes the `reply_to_id` (thread ID)... | open | null | false | 2 | [] | [] | 2026-03-12T19:34:10Z | 2026-05-04T12:47:06Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | fc4ngftsdj-hash | 263,761,580 | U_kgDOD7iurA | User | false |
openclaw/openclaw | 4,066,893,819 | I_kwDOQb6kR87yZ9_7 | 44,353 | https://github.com/openclaw/openclaw/issues/44353 | https://api.github.com/repos/openclaw/openclaw/issues/44353 | Fallback models not triggered on provider-level errors (e.g., AWS Bedrock model ID changes) | ## Summary
Model fallbacks configured in `openclaw.json` are not triggered when the primary model fails with a provider-level error. The error is treated as terminal, leaving the session/subagent failed rather than retrying with a fallback.
## Context
AWS Bedrock recently changed behavior: raw model IDs (`anthropic.... | open | null | false | 5 | [] | [] | 2026-03-12T19:53:09Z | 2026-05-04T12:47:09Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | lucca-alma | 258,741,130 | U_kgDOD2wTig | User | false |
openclaw/openclaw | 4,066,903,951 | I_kwDOQb6kR87yaAeP | 44,354 | https://github.com/openclaw/openclaw/issues/44354 | https://api.github.com/repos/openclaw/openclaw/issues/44354 | Bug: openclaw health --json reports stale Discord state | **Environment:**
- OpenClaw version: 2026.3.11
- OS: Ubuntu (Linux 6.17.0)
- Installation method: curl install script
**Description:**
codeopenclaw health --json reports Discord as running: false and tokenSource: none even when Discord is actively connected and receiving messages.
**Steps to reproduce:**
1. Ensure D... | closed | completed | true | 5 | [] | [] | 2026-03-12T19:55:08Z | 2026-05-02T04:27:53Z | 2026-04-29T15:27:41Z | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | ferenc-acs | 18,510,685 | MDQ6VXNlcjE4NTEwNjg1 | User | false |
openclaw/openclaw | 4,066,970,558 | I_kwDOQb6kR87yaQu- | 44,358 | https://github.com/openclaw/openclaw/issues/44358 | https://api.github.com/repos/openclaw/openclaw/issues/44358 | [Bug]: openprose: agent.model is not always honored | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
I have a 4 agent prose file. Each agent is assigned a different model, e.g.:
```
agent agent_a:
model: openai/gpt-5.2
```
My first run on 3/10 was successful. Each agent ran with the model specified. However in another run I just did ea... | open | null | false | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-03-12T20:10:03Z | 2026-05-04T12:47:11Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | jingxue | 3,587,672 | MDQ6VXNlcjM1ODc2NzI= | User | false |
openclaw/openclaw | 4,067,125,354 | I_kwDOQb6kR87ya2hq | 44,375 | https://github.com/openclaw/openclaw/issues/44375 | https://api.github.com/repos/openclaw/openclaw/issues/44375 | Adding ACP agent to agents.list silently hijacks all routing from implicit main agent | ## Summary
Adding a single ACP-bound agent to `agents.list` (e.g. for a persistent Codex binding on a Telegram topic) silently takes over as the default agent for **all** traffic — DMs, every group topic, heartbeats, crons — replacing the implicit `main` agent. No warning is shown. The user has no indication that thei... | open | null | false | 3 | [] | [] | 2026-03-12T20:39:01Z | 2026-05-04T12:47:16Z | null | CONTRIBUTOR | null | 20260504T180039Z | 2026-05-04T18:00:39Z | barronlroth | 5,567,884 | MDQ6VXNlcjU1Njc4ODQ= | User | false |
openclaw/openclaw | 4,067,768,260 | I_kwDOQb6kR87ydTfE | 44,428 | https://github.com/openclaw/openclaw/issues/44428 | https://api.github.com/repos/openclaw/openclaw/issues/44428 | [Bug]: Isolated cron with delivery.bestEffort:true blocks for full timeout when fire-and-forget subagents are active | **Bug type:** Behavior bug (incorrect output/state without crash)
**Summary:** An isolated cron job with `delivery.bestEffort: true` blocks for the full `timeoutMs` and records a timeout error when the job spawns fire-and-forget subagents, even though the job's own work completed successfully.
**Steps to reproduce:**... | open | null | false | 2 | [] | [] | 2026-03-12T23:01:09Z | 2026-05-04T12:47:20Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | spectra-the-bot | 262,214,320 | U_kgDOD6ESsA | User | false |
openclaw/openclaw | 4,067,787,571 | I_kwDOQb6kR87ydYMz | 44,431 | https://github.com/openclaw/openclaw/issues/44431 | https://api.github.com/repos/openclaw/openclaw/issues/44431 | Browser tool: 7 improvements from real-world automation field test | # Browser Tool Improvements — Field Report
Based on extensive real-world browser automation (email provider signups across 9+ providers) on 2026-03-12.
## Issues Encountered & Proposed Fixes
### 1. 🔴 No CSS Selector Support — Forces Verbose Snapshot→Ref Workflow
**Problem:** Every single interaction requires a sna... | open | null | false | 4 | [] | [] | 2026-03-12T23:06:15Z | 2026-05-04T00:11:41Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | ibadukefan | 3,534,006 | MDQ6VXNlcjM1MzQwMDY= | User | false |
openclaw/openclaw | 4,067,889,429 | I_kwDOQb6kR87ydxEV | 44,448 | https://github.com/openclaw/openclaw/issues/44448 | https://api.github.com/repos/openclaw/openclaw/issues/44448 | [Bug] v2026.3.11: CLI 'openclaw agent --channel discord --to' routes to agent:*:main instead of live channel session | ## Bug Summary
In OpenClaw v2026.3.11, the CLI `openclaw agent --channel discord --to ...` incorrectly resolves the `agent:*:main` session instead of the live session associated with the specified Discord channel/thread.
## Observed Symptoms
- `openclaw agent --channel discord --to ...` appears to deliver but resolv... | open | null | false | 5 | [] | [] | 2026-03-12T23:40:37Z | 2026-05-04T12:47:22Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | Mdx2025 | 181,967,107 | U_kgDOCtiZAw | User | false |
openclaw/openclaw | 4,067,940,229 | I_kwDOQb6kR87yd9eF | 44,460 | https://github.com/openclaw/openclaw/issues/44460 | https://api.github.com/repos/openclaw/openclaw/issues/44460 | doctor --fix leaves legacy notify:true cron field unresolved when cron.webhook is unset | ## Description
`openclaw doctor --fix` does not fully migrate legacy cron metadata when a job still has top-level `notify: true` and `cron.webhook` is unset.
Observed behavior:
- Doctor reports legacy cron storage and warns about `notify: true` fallback.
- Running `openclaw doctor --fix` prints `Cron store normalize... | open | null | false | 3 | [] | [] | 2026-03-12T23:58:04Z | 2026-05-04T12:47:23Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | kai-the-man[bot] | 259,478,402 | BOT_kgDOD3dTgg | Bot | false |
openclaw/openclaw | 4,068,004,052 | I_kwDOQb6kR87yeNDU | 44,471 | https://github.com/openclaw/openclaw/issues/44471 | https://api.github.com/repos/openclaw/openclaw/issues/44471 | [Bug] Runtime event wording causes premature success claims in subagent flows | ## Problem
The runtime currently emits an internal completion event with wording "ready for user delivery" which agents interpret as a signal that the task is complete and ready to be reported to the user.
However, in subagent workflows, this creates a critical bug where:
- "subagent completed successfully" is treate... | open | null | false | 2 | [] | [] | 2026-03-13T00:17:47Z | 2026-05-04T12:47:25Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | Mdx2025 | 181,967,107 | U_kgDOCtiZAw | User | false |
openclaw/openclaw | 4,068,064,511 | I_kwDOQb6kR87yebz_ | 44,484 | https://github.com/openclaw/openclaw/issues/44484 | https://api.github.com/repos/openclaw/openclaw/issues/44484 | Agent declared tools != effective session tools: no strict inheritance mode, no visibility, no clipping reason trace | ### Summary
OpenClaw currently allows configuring tool permissions at the agent level (for example `agents.list[].tools.profile = "full"` or explicit allow/deny rules), but in practice the tools available to a concrete runtime session can still be reduced by subagent/session/runtime-specific behavior.
This creates a ... | open | null | false | 2 | [] | [] | 2026-03-13T00:38:50Z | 2026-05-04T12:47:27Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | jayv29 | 260,694,162 | U_kgDOD4ngkg | User | false |
openclaw/openclaw | 4,068,182,362 | I_kwDOQb6kR87ye4la | 44,502 | https://github.com/openclaw/openclaw/issues/44502 | https://api.github.com/repos/openclaw/openclaw/issues/44502 | [Bug]: Discord routing / mention-gating issue in OpenClaw | ### Bug type
Regression (worked before, now fails)
### Summary
## Summary
I found a Discord preflight / mention-gating bug where message handling is too
permissive in some agent-routing cases.
The issue is not just “mention required vs not required”. The preflight logic
can still treat the wrong content as... | open | null | false | 5 | [
"bug",
"regression"
] | [] | 2026-03-13T01:14:48Z | 2026-05-04T12:47:29Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | JustAskNudge | 259,488,613 | U_kgDOD3d7ZQ | User | false |
openclaw/openclaw | 4,068,301,592 | I_kwDOQb6kR87yfVsY | 44,534 | https://github.com/openclaw/openclaw/issues/44534 | https://api.github.com/repos/openclaw/openclaw/issues/44534 | Heartbeat skipped reason 'target-none' is misleading when delivery is disabled by default | ## Summary
When heartbeats run with the default configuration, `openclaw system heartbeat last` (and gateway logs) can show:
```json
{ "status": "skipped", "reason": "target-none" }
```
This is accurate (the default heartbeat delivery target is `none`), but the wording is misleading: it reads like the system failed t... | open | null | false | 2 | [] | [] | 2026-03-13T01:59:47Z | 2026-05-04T13:18:36Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | Congregalis | 25,055,688 | MDQ6VXNlcjI1MDU1Njg4 | User | false |
openclaw/openclaw | 4,068,340,384 | I_kwDOQb6kR87yffKg | 44,540 | https://github.com/openclaw/openclaw/issues/44540 | https://api.github.com/repos/openclaw/openclaw/issues/44540 | Hybrid search BM25 component penalizes multimodal (image/audio) results | ## Summary
When `memorySearch.multimodal.enabled = true` with Gemini embedding 2, image and audio files are properly indexed with valid embeddings. However, they never surface in `memory_search` results under the default hybrid search configuration.
## Root Cause
Hybrid search computes: `finalScore = vectorWeight × ... | open | null | false | 2 | [] | [] | 2026-03-13T02:12:44Z | 2026-05-04T13:18:38Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | markclawbot | 258,504,883 | U_kgDOD2h4sw | User | false |
openclaw/openclaw | 4,068,415,392 | I_kwDOQb6kR87yfxeg | 44,559 | https://github.com/openclaw/openclaw/issues/44559 | https://api.github.com/repos/openclaw/openclaw/issues/44559 | [Bug]: Windows： Gateway 关闭 PowerShell 窗口后断连 | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
在 Windows 上使用 `openclaw gateway install` 安装的计划任务，在关闭 PowerShell 窗口后 Gateway 会断连，必须重新运行 `openclaw gateway start` 才能恢复。
### Steps to reproduce
1. `openclaw gateway install` - 计划任务使用 InteractiveToken，依赖用户会话
2. 手动用 `schtasks /Create /TN "Ope... | open | null | false | 4 | [
"bug",
"bug:behavior"
] | [] | 2026-03-13T02:37:09Z | 2026-05-04T13:18:41Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | 781791138-bot | 267,816,328 | U_kgDOD_aNiA | User | false |
openclaw/openclaw | 4,068,472,344 | I_kwDOQb6kR87yf_YY | 44,571 | https://github.com/openclaw/openclaw/issues/44571 | https://api.github.com/repos/openclaw/openclaw/issues/44571 | [Bug]: onboard writes auth-profiles only to agents/main/, not to all configured agents | ## Summary
When running `openclaw onboard` with multiple agents configured in `openclaw.json`, the auth token (or tokenRef) is only written to `~/.openclaw/agents/main/agent/auth-profiles.json`. It is **not distributed** to the other configured agents (e.g., `agents/chief/`, `agents/developer/`, `agents/qa/`, etc.).
... | open | null | false | 2 | [] | [] | 2026-03-13T02:54:22Z | 2026-05-04T13:18:42Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | wenboown | 1,467,314 | MDQ6VXNlcjE0NjczMTQ= | User | false |
openclaw/openclaw | 4,068,526,955 | I_kwDOQb6kR87ygMtr | 44,584 | https://github.com/openclaw/openclaw/issues/44584 | https://api.github.com/repos/openclaw/openclaw/issues/44584 | [Bug]: Discord threads - volatile metadata re-injection + CLI routing to wrong session | ## Problem
Two related regressions affecting live Discord sessions in OpenClaw:
1. **Volatile metadata re-injected every turn**: In Discord threads, `message_id`, `reply_to_id`, `timestamp`, `Thread starter`, and `Chat history since last reply` were being re-injected on every turn, causing progressive context degrada... | open | null | false | 3 | [] | [] | 2026-03-13T03:09:22Z | 2026-05-04T00:05:40Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | Mdx2025 | 181,967,107 | U_kgDOCtiZAw | User | false |
openclaw/openclaw | 4,068,574,982 | I_kwDOQb6kR87ygYcG | 44,599 | https://github.com/openclaw/openclaw/issues/44599 | https://api.github.com/repos/openclaw/openclaw/issues/44599 | [Bug]: OPENCLAW_CONFIG_DIR cannot contain whitespace | ### Bug type
Regression (worked before, now fails)
### Summary
I tried to install openclaw with docker as this instruction (https://www.youtube.com/watch?v=V8nKksknXxg) but faced the error: OPENCLAW_CONFIG_DIR cannot contain whitespace when running `.\docker-setup.sh`.
### Steps to reproduce
1. git clone https://g... | open | null | false | 6 | [
"bug",
"regression"
] | [] | 2026-03-13T03:23:13Z | 2026-05-04T00:11:42Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | mainguyenanhvu | 26,211,591 | MDQ6VXNlcjI2MjExNTkx | User | false |
openclaw/openclaw | 4,068,672,512 | I_kwDOQb6kR87ygwQA | 44,635 | https://github.com/openclaw/openclaw/issues/44635 | https://api.github.com/repos/openclaw/openclaw/issues/44635 | Feature Request: Per-agent browser profile isolation (userDataDir) | Hello OpenClaw team!
I'm trying to configure independent browser storage (cookies, localStorage, etc.) for different agents to prevent them from interfering with each other.
**Use Case:**
- Multiple agents need to log into different accounts (e.g., different admin systems)
- Each agent should have isolated browser id... | closed | not_planned | false | 5 | [] | [] | 2026-03-13T03:56:22Z | 2026-05-04T01:08:28Z | 2026-05-04T01:08:28Z | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | tom0932 | 16,838,887 | MDQ6VXNlcjE2ODM4ODg3 | User | false |
openclaw/openclaw | 4,070,127,641 | I_kwDOQb6kR87ymTgZ | 44,965 | https://github.com/openclaw/openclaw/issues/44965 | https://api.github.com/repos/openclaw/openclaw/issues/44965 | Feature Request: Stream Repetition Safeguard (Halt & Confirm) | ### Is your feature request related to a problem? Please describe.
When using certain models, the generation can occasionally get stuck in an infinite loop, spamming the exact same phrase (e.g., "Done, I will output it") endlessly. This floods chat channels with identical lines of text and wastes tokens.
### Describe ... | open | null | false | 2 | [] | [] | 2026-03-13T10:06:39Z | 2026-05-04T00:05:55Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | garysui | 8,386,050 | MDQ6VXNlcjgzODYwNTA= | User | false |
openclaw/openclaw | 4,072,299,548 | I_kwDOQb6kR87yulwc | 45,314 | https://github.com/openclaw/openclaw/issues/45314 | https://api.github.com/repos/openclaw/openclaw/issues/45314 | [Bug]: Early abort response templates are not populated. | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Template variables are not populated when user issues early stop commands.
### Steps to reproduce
1. Configure a `responsePrefix` with `{model}` and/or `{thinkingLevel}` template variables
2. Start a conversation so the agent is processi... | open | null | false | 7 | [
"bug",
"bug:behavior"
] | [] | 2026-03-13T17:00:13Z | 2026-05-04T00:12:05Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | christopherwoodall | 363,708 | MDQ6VXNlcjM2MzcwOA== | User | false |
openclaw/openclaw | 4,072,375,827 | I_kwDOQb6kR87yu4YT | 45,323 | https://github.com/openclaw/openclaw/issues/45323 | https://api.github.com/repos/openclaw/openclaw/issues/45323 | Feature Request: Slack-Style @Mention Autocomplete in Control UI Chat | ## Summary
Add Slack-style `@agent-name` mention autocomplete to the Control UI chat input. When a user types `@`, a dropdown directory of known agents appears, filtered as they type. Selecting an agent inserts a styled, highlighted mention chip into the message — visually distinct from plain text, exactly like Slack'... | open | null | false | 2 | [] | [] | 2026-03-13T17:14:32Z | 2026-05-04T00:12:07Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | clawtomationagent | 267,995,670 | U_kgDOD_lKFg | User | false |
openclaw/openclaw | 4,072,395,708 | I_kwDOQb6kR87yu9O8 | 45,326 | https://github.com/openclaw/openclaw/issues/45326 | https://api.github.com/repos/openclaw/openclaw/issues/45326 | [Bug]: TUI: Input typed during model generation is swallowed and incorrectly queued for the next turn (Interrupt failure) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When using openclaw tui, attempting to interrupt the AI while it is actively generating a response (or thinking) fails. Any text typed during the generation process is visually "swallowed" and ignored. However, this swallowed input is impro... | open | null | false | 10 | [
"bug",
"bug:behavior"
] | [] | 2026-03-13T17:18:29Z | 2026-05-04T13:19:16Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | Jinshi-Liu | 182,833,437 | U_kgDOCuXRHQ | User | false |
openclaw/openclaw | 4,072,668,651 | I_kwDOQb6kR87yv_3r | 45,359 | https://github.com/openclaw/openclaw/issues/45359 | https://api.github.com/repos/openclaw/openclaw/issues/45359 | Intermittent server_error on GPT-5.4 after OAuth refresh; workaround switch to 5.3 then back | **Описание**
При работе через OpenClaw на модели `openai-codex/gpt-5.4` периодически возникает ошибка:
```
Codex error: {"type":"error","error":{"type":"server_error","code":"server_error","message":"An error occurred while processing your request... request ID <пример>"}}
```
Ошибка чаще проявляется после того, как ... | open | null | false | 3 | [] | [] | 2026-03-13T18:11:34Z | 2026-05-04T00:12:08Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | IIIyban | 137,302,231 | U_kgDOCC8Q1w | User | false |
openclaw/openclaw | 4,072,800,825 | I_kwDOQb6kR87ywgI5 | 45,382 | https://github.com/openclaw/openclaw/issues/45382 | https://api.github.com/repos/openclaw/openclaw/issues/45382 | [Feature]: add VALUE.md as a first-class workspace context file | ## Summary
Add a new default workspace context file, `VALUE.md`, as a first-class bootstrap/context file.
## Problem to solve
OpenClaw already has strong workspace concepts for agent behavior and continuity, including `AGENTS.md`, `SOUL.md`, `USER.md`, `TOOLS.md`, and optional `MEMORY.md`.
Today, `SOUL.md` is the c... | open | null | false | 2 | [] | [] | 2026-03-13T18:41:52Z | 2026-05-04T00:12:11Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | ZhanKunLiAuto | 90,255,868 | MDQ6VXNlcjkwMjU1ODY4 | User | false |
openclaw/openclaw | 4,072,839,289 | I_kwDOQb6kR87ywph5 | 45,385 | https://github.com/openclaw/openclaw/issues/45385 | https://api.github.com/repos/openclaw/openclaw/issues/45385 | Raw API error leaks to user: tool_use/tool_result mismatch not caught by formatAssistantErrorText | ## Bug Description
When a tool call result is dropped (timeout, race condition, etc.), the Anthropic API rejects the conversation with an `invalid_request_error` about `tool_use` IDs found without `tool_result` blocks. This raw error message is surfaced directly to the user instead of being rewritten to a friendly mes... | open | null | false | 3 | [] | [] | 2026-03-13T18:51:40Z | 2026-05-04T00:12:12Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | thelaunchpadtlp | 19,968,575 | MDQ6VXNlcjE5OTY4NTc1 | User | false |
openclaw/openclaw | 4,072,875,670 | I_kwDOQb6kR87ywyaW | 45,388 | https://github.com/openclaw/openclaw/issues/45388 | https://api.github.com/repos/openclaw/openclaw/issues/45388 | TUI --session mode doesn't live-stream messages | ## Bug
When running:
```
openclaw tui --session "telegram:direct:<id>" --deliver
```
The TUI loads message history on launch but does **not** receive real-time updates.
### Observed behavior
- New inbound Telegram messages don't appear in the TUI
- Agent replies don't echo back into the TUI view
- The only way to se... | open | null | false | 5 | [] | [] | 2026-03-13T18:59:47Z | 2026-05-04T00:52:22Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | bwatchmitch | 259,208,597 | U_kgDOD3M1lQ | User | false |
openclaw/openclaw | 4,072,877,503 | I_kwDOQb6kR87ywy2_ | 45,389 | https://github.com/openclaw/openclaw/issues/45389 | https://api.github.com/repos/openclaw/openclaw/issues/45389 | Feature request: Circuit breaker for consecutive LLM timeouts | ## Problem
When an LLM request times out, OpenClaw retries on the next heartbeat/message. If the root cause persists (e.g., context too large, provider degradation), the system retries indefinitely — compounding the failure. We observed 233 timeouts across 13 agents over 10 days, with one agent stuck in a retry loop f... | closed | not_planned | true | 3 | [] | [] | 2026-03-13T19:00:18Z | 2026-05-05T04:27:32Z | 2026-05-02T11:24:18Z | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | pescaohq-web | 257,331,994 | U_kgDOD1aTGg | User | false |
openclaw/openclaw | 4,072,878,201 | I_kwDOQb6kR87ywzB5 | 45,390 | https://github.com/openclaw/openclaw/issues/45390 | https://api.github.com/repos/openclaw/openclaw/issues/45390 | Feature request: Session TTL / max lifetime for automatic rotation | ## Problem
Agent sessions accumulate context indefinitely. Without a maximum lifetime, sessions grow until they hit provider limits or cause timeouts. We observed sessions running for 6+ days without rotation, accumulating 171k/200k tokens and 71 timeouts.
## Proposed Solution
Add a configurable session TTL:
```jso... | open | null | false | 3 | [] | [] | 2026-03-13T19:00:29Z | 2026-05-04T13:19:18Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | pescaohq-web | 257,331,994 | U_kgDOD1aTGg | User | false |
openclaw/openclaw | 4,072,965,243 | I_kwDOQb6kR87yxIR7 | 45,404 | https://github.com/openclaw/openclaw/issues/45404 | https://api.github.com/repos/openclaw/openclaw/issues/45404 | Low-priority: detect and clean up legacy clawd managed Brave profile residue after rename migration | ## Summary
Rename-era installs can leave a duplicate legacy managed Brave profile on disk under a `clawd` path even though the canonical managed browser profile is now `openclaw`.
## Observed behavior
On one migrated local setup, both of these existed:
- `~/.openclaw/browser/openclaw/user-data`
- `~/.openclaw/brows... | open | null | false | 3 | [] | [] | 2026-03-13T19:20:09Z | 2026-05-04T00:12:15Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | mickobizzle | 182,108,048 | U_kgDOCtq_kA | User | false |
openclaw/openclaw | 4,073,055,161 | I_kwDOQb6kR87yxeO5 | 45,415 | https://github.com/openclaw/openclaw/issues/45415 | https://api.github.com/repos/openclaw/openclaw/issues/45415 | Feature request: MEMORY.md size warning/limit enforcement | ## Problem
MEMORY.md is silently truncated at ~20K characters. Users have no warning when approaching this limit, leading to lost context.
## Current state
Our MEMORY.md is at 19,770 chars — 230 chars from silent truncation. We only discovered this by manually checking `wc -c`.
## Proposed Solution
1. **Size indic... | open | null | false | 3 | [] | [] | 2026-03-13T19:41:30Z | 2026-05-04T00:12:16Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | sabo961 | 191,984 | MDQ6VXNlcjE5MTk4NA== | User | false |
openclaw/openclaw | 4,073,520,831 | I_kwDOQb6kR87yzP6_ | 45,475 | https://github.com/openclaw/openclaw/issues/45475 | https://api.github.com/repos/openclaw/openclaw/issues/45475 | Feature: channels.slack.ignoreSubtypes config key to suppress message_changed / message_deleted events | ## Problem
When a Slack user edits or deletes a message, Slack fires `message_changed` / `message_deleted` events. OpenClaw converts these to system events which can wake an agent session and trigger a response — even though no new user input was provided.
Currently there is no config key to suppress these subtypes a... | open | null | false | 3 | [] | [] | 2026-03-13T21:41:16Z | 2026-05-04T13:19:20Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | ctcreel | 1,657,759 | MDQ6VXNlcjE2NTc3NTk= | User | false |
openclaw/openclaw | 4,073,623,615 | I_kwDOQb6kR87yzpA_ | 45,488 | https://github.com/openclaw/openclaw/issues/45488 | https://api.github.com/repos/openclaw/openclaw/issues/45488 | [Bug]: #18049 introduces session bloat | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
#18049 introduced a functionality that copies parts of the system prompt into the context after compaction. This, however, is completely useless as the system prompt is not compacted away.
This leads to (a) session bloat, and (b) "magic" b... | open | null | false | 5 | [
"bug",
"bug:behavior"
] | [] | 2026-03-13T22:11:32Z | 2026-05-04T00:12:28Z | null | CONTRIBUTOR | null | 20260504T030031Z | 2026-05-04T03:00:31Z | HenryLoenwind | 1,485,873 | MDQ6VXNlcjE0ODU4NzM= | User | false |
openclaw/openclaw | 4,073,652,012 | I_kwDOQb6kR87yzv8s | 45,494 | https://github.com/openclaw/openclaw/issues/45494 | https://api.github.com/repos/openclaw/openclaw/issues/45494 | [Bug]: Cron agent jobs silently time out during sustained LLM API outages instead of fast-failing on definitive errors | ### Bug type
Regression (worked before, now fails)
### Summary
When the primary LLM provider returns sustained HTTP 500 (Internal Server Error) responses, OpenClaw cron agent jobs exhaust their full `timeoutSeconds` window (e.g. 180s) on each attempt rather than fast-failing on a definitive API error. This results i... | open | null | false | 6 | [
"bug",
"regression"
] | [] | 2026-03-13T22:18:55Z | 2026-05-04T13:19:21Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | 03marcbluechain | 142,078,325 | U_kgDOCHfxdQ | User | false |
openclaw/openclaw | 4,073,692,264 | I_kwDOQb6kR87yz5xo | 45,501 | https://github.com/openclaw/openclaw/issues/45501 | https://api.github.com/repos/openclaw/openclaw/issues/45501 | Feature: `session.resetPrompt` — configurable session startup message | ## Problem
When a session resets via `/new` or `/reset`, OpenClaw injects a hardcoded startup message into the conversation:
> *"A new session was started via /new or /reset. Execute your Session Startup sequence now - read the required files before responding to the user..."*
This text is great for bare/unconfigure... | open | null | false | 5 | [] | [] | 2026-03-13T22:30:39Z | 2026-05-04T13:19:23Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | SwivelLabs | 261,235,255 | U_kgDOD5IiNw | User | false |
openclaw/openclaw | 4,073,711,188 | I_kwDOQb6kR87yz-ZU | 45,503 | https://github.com/openclaw/openclaw/issues/45503 | https://api.github.com/repos/openclaw/openclaw/issues/45503 | Manual context clearing for tool results | Currently, tool results are only pruned from context via automatic TTL (e.g., 1h with `cache-ttl` mode). This creates inefficiency when large tool results (emails, search results, etc.) are only needed momentarily.
**Use case:**
1. User asks: "Find my first iPhone purchase in Gmail"
2. Agent fetches 25 emails (80k+ to... | open | null | false | 3 | [] | [] | 2026-03-13T22:35:55Z | 2026-05-04T00:12:29Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | tbertran | 2,054,914 | MDQ6VXNlcjIwNTQ5MTQ= | User | false |
openclaw/openclaw | 4,073,717,104 | I_kwDOQb6kR87yz_1w | 45,505 | https://github.com/openclaw/openclaw/issues/45505 | https://api.github.com/repos/openclaw/openclaw/issues/45505 | Generalize post-timeout compaction completion reconciliation across session state | ## Summary
OpenClaw now has a narrow fix for one specific state mismatch: compaction can succeed after the runner stops waiting, and `sessions.json.compactionCount` needs to be reconciled afterward.
That solves a real bug, but it likely points to a broader design gap:
- compaction completion can become known **after... | open | null | false | 3 | [] | [] | 2026-03-13T22:38:15Z | 2026-05-04T13:19:24Z | null | CONTRIBUTOR | null | 20260504T180039Z | 2026-05-04T18:00:39Z | jackal092927 | 3,854,860 | MDQ6VXNlcjM4NTQ4NjA= | User | false |
openclaw/openclaw | 4,074,028,381 | I_kwDOQb6kR87y1L1d | 45,550 | https://github.com/openclaw/openclaw/issues/45550 | https://api.github.com/repos/openclaw/openclaw/issues/45550 | feat(anthropic): migrate 1M context from beta to GA | ## Summary
Anthropic has officially moved the 1M context window from Beta to GA (General Availability). This issue tracks the migration to align OpenClaw with the new Anthropic API behavior.
## Background
Per Anthropic's announcement (https://www.cnbeta.com.tw/articles/tech/1553412.htm), the following changes are no... | open | null | false | 8 | [] | [] | 2026-03-14T00:19:06Z | 2026-05-04T13:19:26Z | null | CONTRIBUTOR | null | 20260504T180039Z | 2026-05-04T18:00:39Z | JonathanJing | 17,068,507 | MDQ6VXNlcjE3MDY4NTA3 | User | false |
openclaw/openclaw | 4,074,069,962 | I_kwDOQb6kR87y1V_K | 45,554 | https://github.com/openclaw/openclaw/issues/45554 | https://api.github.com/repos/openclaw/openclaw/issues/45554 | [Bug]: switching session.dmScope back to main does not reconcile stale peer-keyed direct DM sessions | ## Summary
Changing DM routing back to `session.dmScope = "main"` did **not** reconcile or retire old peer-keyed direct sessions on this machine.
That left multiple live direct-session identities coexisting, including the canonical main session and stale Telegram-peer-keyed direct sessions. In practice this created m... | open | null | false | 4 | [] | [] | 2026-03-14T00:35:15Z | 2026-05-04T05:46:44Z | null | NONE | null | 20260504T060031Z | 2026-05-04T06:00:31Z | SatsConvert | 212,145,289 | U_kgDODKUUiQ | User | false |
openclaw/openclaw | 4,074,094,401 | I_kwDOQb6kR87y1b9B | 45,564 | https://github.com/openclaw/openclaw/issues/45564 | https://api.github.com/repos/openclaw/openclaw/issues/45564 | Add confirmation step to /new and /reset to prevent accidental session wipes | Currently `/new` and `/reset` immediately wipe the session with no confirmation. It's easy to accidentally trigger this and lose your conversation history.
**Request:** Add a multi-step confirmation (e.g., requiring 2-3 clicks/confirmations) before wiping a session. This would prevent accidental data loss, especially ... | open | null | false | 4 | [] | [] | 2026-03-14T00:43:53Z | 2026-05-04T13:19:27Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | nikHowlett | 13,320,810 | MDQ6VXNlcjEzMzIwODEw | User | false |
openclaw/openclaw | 4,074,097,937 | I_kwDOQb6kR87y1c0R | 45,565 | https://github.com/openclaw/openclaw/issues/45565 | https://api.github.com/repos/openclaw/openclaw/issues/45565 | Feature: config option to route gateway lifecycle warnings to a dedicated channel | ## Feature Request
**Problem:**
When the gateway emits lifecycle warnings (e.g. `memorySearch failed`, provider init failures), they are posted to whichever Discord channel is currently active. This creates noise in conversation channels unrelated to system health.
**Example:**
After a gateway restart, the Gemini emb... | open | null | false | 6 | [] | [] | 2026-03-14T00:45:04Z | 2026-05-04T13:19:29Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | Stache73 | 7,196,384 | MDQ6VXNlcjcxOTYzODQ= | User | false |
openclaw/openclaw | 4,074,109,721 | I_kwDOQb6kR87y1fsZ | 45,569 | https://github.com/openclaw/openclaw/issues/45569 | https://api.github.com/repos/openclaw/openclaw/issues/45569 | [Bug]: /status in the web ui needs <enter> pressed twice | ### Bug type
Regression (worked before, now fails)
### Summary
After upgrading to 2026.3.12, I now have to press enter twice to execute /status.
Also, the history cut short is very annoying as it also counts tool calls... (just a remark, not part of the issue)
### Steps to reproduce
1. type "/", "s", "t", "a", "t... | open | null | false | 3 | [
"bug",
"regression"
] | [] | 2026-03-14T00:49:02Z | 2026-05-04T13:19:30Z | null | CONTRIBUTOR | null | 20260504T180039Z | 2026-05-04T18:00:39Z | HenryLoenwind | 1,485,873 | MDQ6VXNlcjE0ODU4NzM= | User | false |
openclaw/openclaw | 4,074,131,680 | I_kwDOQb6kR87y1lDg | 45,573 | https://github.com/openclaw/openclaw/issues/45573 | https://api.github.com/repos/openclaw/openclaw/issues/45573 | Group chat sessions not persisted — only 1 session from 166+ messages | ## Summary
OpenClaw bot receives and responds to group chat messages, but sessions are not being persisted. A group with 166+ messages (including bot responses) only has 1 session recorded via `paw-monitor session list`.
## Environment
- OpenClaw version: 2026.3.8
- Platform: macOS (Apple Silicon, arm64)
- Gateway: ... | open | null | false | 4 | [] | [] | 2026-03-14T00:56:41Z | 2026-05-04T13:19:32Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | kiki830621 | 17,489,494 | MDQ6VXNlcjE3NDg5NDk0 | User | false |
openclaw/openclaw | 4,074,134,594 | I_kwDOQb6kR87y1lxC | 45,574 | https://github.com/openclaw/openclaw/issues/45574 | https://api.github.com/repos/openclaw/openclaw/issues/45574 | Slack download-file action: fileId param not passed through message tool schema | ## Bug
The Slack `download-file` action is registered and the handler works, but the `fileId` parameter required by the Slack handler (`readStringParam(params, "fileId", { required: true })`) is not included in the message tool's parameter schema.
This means when an agent calls `message(action="download-file", channe... | closed | completed | true | 4 | [] | [] | 2026-03-14T00:57:43Z | 2026-05-03T04:37:09Z | 2026-04-30T14:06:17Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | chadvegas | 150,620,074 | U_kgDOCPpHqg | User | false |
openclaw/openclaw | 4,074,230,940 | I_kwDOQb6kR87y19Sc | 45,595 | https://github.com/openclaw/openclaw/issues/45595 | https://api.github.com/repos/openclaw/openclaw/issues/45595 | [Bug]: Exec approval checks use symlink path but execution uses real path | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
The exec approval system checks `resolvedPath` (may be a symlink) but execution pins to `resolvedRealPath` (the real file). The system computes both paths but uses different ones for trust vs execution - so it can approve or reject one path... | open | null | false | 3 | [] | [] | 2026-03-14T01:26:53Z | 2026-05-04T00:07:15Z | null | CONTRIBUTOR | null | 20260504T030031Z | 2026-05-04T03:00:31Z | jasonftl | 175,325,999 | U_kgDOCnNDLw | User | false |
openclaw/openclaw | 4,074,270,191 | I_kwDOQb6kR87y2G3v | 45,608 | https://github.com/openclaw/openclaw/issues/45608 | https://api.github.com/repos/openclaw/openclaw/issues/45608 | [Feature]: Pre-reset agentic memory flush — /new and daily reset should get the same memory flush as compaction | ### Summary
Run a silent agentic memory flush before `/new`, `/reset`, and the daily reset destroy the session, using the same mechanism that already runs before compaction.
### Problem to solve
When a session nears compaction, OpenClaw runs a silent agentic turn (`runMemoryFlushIfNeeded`) where the agent writes cur... | open | null | false | 8 | [] | [] | 2026-03-14T01:44:01Z | 2026-05-04T00:12:41Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | kamikariat | 52,008,157 | MDQ6VXNlcjUyMDA4MTU3 | User | false |
openclaw/openclaw | 4,074,343,512 | I_kwDOQb6kR87y2YxY | 45,638 | https://github.com/openclaw/openclaw/issues/45638 | https://api.github.com/repos/openclaw/openclaw/issues/45638 | fix(telegram): callback_query passes wrong messageIdOverride (callback.id instead of message_id) | ## Bug Description
In `src/telegram/bot-handlers.ts`, when a Telegram inline button callback reaches `processMessage`, it passes `callback.id` (the callback query's unique UUID) as `messageIdOverride` instead of `callbackMessage.message_id` (the actual Telegram message ID).
**File/line:** `src/telegram/bot-handlers.t... | open | null | false | 2 | [] | [] | 2026-03-14T02:15:03Z | 2026-05-04T00:12:43Z | null | CONTRIBUTOR | null | 20260504T030031Z | 2026-05-04T03:00:31Z | bboyyan | 202,459,156 | U_kgDODBFIFA | User | false |
openclaw/openclaw | 4,074,369,026 | I_kwDOQb6kR87y2fAC | 45,649 | https://github.com/openclaw/openclaw/issues/45649 | https://api.github.com/repos/openclaw/openclaw/issues/45649 | [Bug]: Tool bubble is unreadable in v2 UI | ### Bug type
Regression (worked before, now fails)
### Summary
<img width="1325" height="290" alt="Image" src="https://github.com/user-attachments/assets/e038ea86-c868-427e-bbbd-3a02758ebe67" />
medium grey on light grey. I can't read a word. Also, why is it cut off so early? there's more than enough space for the ... | open | null | false | 5 | [
"bug",
"regression"
] | [] | 2026-03-14T02:24:39Z | 2026-05-04T13:19:34Z | null | CONTRIBUTOR | null | 20260504T180039Z | 2026-05-04T18:00:39Z | HenryLoenwind | 1,485,873 | MDQ6VXNlcjE0ODU4NzM= | User | false |
openclaw/openclaw | 4,074,395,266 | I_kwDOQb6kR87y2laC | 45,655 | https://github.com/openclaw/openclaw/issues/45655 | https://api.github.com/repos/openclaw/openclaw/issues/45655 | Bug: Poe image/video-output models are accepted in config but fail at runtime (Nano Banana 400, Veo/Sora fallback) | ## Summary
OpenClaw appears to accept Poe media-generation models in `models.providers` config, but image/video-output models are not handled correctly at runtime.
In my setup, Poe text-output models work, and `poe/gemini-3.1-pro` also works for normal chat / multimodal-understanding flows. But Poe media-output model... | open | null | false | 4 | [] | [] | 2026-03-14T02:34:14Z | 2026-05-04T13:19:35Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | zeming27 | 260,163,139 | U_kgDOD4HGQw | User | false |
openclaw/openclaw | 4,074,397,589 | I_kwDOQb6kR87y2l-V | 45,656 | https://github.com/openclaw/openclaw/issues/45656 | https://api.github.com/repos/openclaw/openclaw/issues/45656 | [Bug]: Chat line has bad usability (v2 UI) | ### Bug type
Regression (worked before, now fails)
### Summary
<img width="3366" height="149" alt="Image" src="https://github.com/user-attachments/assets/17fc872b-3543-49d0-82ac-76c642faa3dd" />
see image. Only the upper half, which has no visual distinction from the lower half, can be clicked. If you miss it (natu... | open | null | false | 3 | [
"bug",
"regression"
] | [] | 2026-03-14T02:35:16Z | 2026-05-04T00:12:45Z | null | CONTRIBUTOR | null | 20260504T030031Z | 2026-05-04T03:00:31Z | HenryLoenwind | 1,485,873 | MDQ6VXNlcjE0ODU4NzM= | User | false |
openclaw/openclaw | 4,074,496,689 | I_kwDOQb6kR87y2-Kx | 45,688 | https://github.com/openclaw/openclaw/issues/45688 | https://api.github.com/repos/openclaw/openclaw/issues/45688 | [Bug]: Config section is horriby convoluted now (v2 UI) | ### Bug type
Regression (worked before, now fails)
### Summary
old:
<img width="428" height="1673" alt="Image" src="https://github.com/user-attachments/assets/fed0a3ee-5935-445c-b248-bd68d03fd823" />
Clearly sorted list, filter, search and mode toggle all in one place.
new:
<img width="310" height="425" alt="Ima... | closed | not_planned | true | 3 | [
"bug",
"regression"
] | [
"BunsDev"
] | 2026-03-14T03:16:59Z | 2026-05-05T04:27:21Z | 2026-05-02T12:16:50Z | CONTRIBUTOR | null | 20260505T060034Z | 2026-05-05T06:00:34Z | HenryLoenwind | 1,485,873 | MDQ6VXNlcjE0ODU4NzM= | User | false |
openclaw/openclaw | 4,074,532,627 | I_kwDOQb6kR87y3G8T | 45,698 | https://github.com/openclaw/openclaw/issues/45698 | https://api.github.com/repos/openclaw/openclaw/issues/45698 | Bug: Control UI becomes progressively stuck after being open for a while (2026.3.12) | # Bug: Control UI becomes progressively stuck after being open for a while (2026.3.12)
## Summary
After upgrading to **OpenClaw 2026.3.12**, the **Control UI / dashboard becomes progressively sluggish and then effectively stuck after being open for a while**.
This is **not limited to the Channels tab**.
At first the... | open | null | false | 7 | [] | [] | 2026-03-14T03:34:17Z | 2026-05-04T00:12:54Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | JLDynamics | 240,741,042 | U_kgDODllqsg | User | false |
openclaw/openclaw | 4,074,546,554 | I_kwDOQb6kR87y3KV6 | 45,703 | https://github.com/openclaw/openclaw/issues/45703 | https://api.github.com/repos/openclaw/openclaw/issues/45703 | [Bug]: Agent section is horrible (v2 UI) | ### Bug type
Regression (worked before, now fails)
### Summary
now:
<img width="2558" height="1013" alt="Image" src="https://github.com/user-attachments/assets/85636714-167c-4ec3-8e9e-63e0e9603121" />
before:
<img width="2760" height="1064" alt="Image" src="https://github.com/user-attachments/assets/68eec8e5-804e... | open | null | false | 5 | [
"bug",
"regression"
] | [] | 2026-03-14T03:42:15Z | 2026-05-04T00:12:55Z | null | CONTRIBUTOR | null | 20260504T030031Z | 2026-05-04T03:00:31Z | HenryLoenwind | 1,485,873 | MDQ6VXNlcjE0ODU4NzM= | User | false |
openclaw/openclaw | 4,074,550,646 | I_kwDOQb6kR87y3LV2 | 45,706 | https://github.com/openclaw/openclaw/issues/45706 | https://api.github.com/repos/openclaw/openclaw/issues/45706 | [Bug]: error=HTTP 422: "Check open ai req parameter error" | ### Bug type
Regression (worked before, now fails)
### Summary
# error=HTTP 422: "Check open ai req parameter error"
在本地部署了deepseek满血版的大模型，使用mindie框架并开放openapi接口服务
调用工具的接口都能正常调用
<img width="1739" height="681" alt="Image" src="https://github.com/user-attachments/assets/6824b639-c81f-42a6-b061-01745fb9ec39" />
... | open | null | false | 6 | [
"bug",
"regression"
] | [] | 2026-03-14T03:44:17Z | 2026-05-04T13:19:36Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | Zha-Miku | 54,501,325 | MDQ6VXNlcjU0NTAxMzI1 | User | false |
openclaw/openclaw | 4,074,558,571 | I_kwDOQb6kR87y3NRr | 45,709 | https://github.com/openclaw/openclaw/issues/45709 | https://api.github.com/repos/openclaw/openclaw/issues/45709 | [Bug]: Usage header is doubled (v2 UI) | ### Bug type
Regression (worked before, now fails)
### Summary
<img width="782" height="499" alt="Image" src="https://github.com/user-attachments/assets/a5839143-55bb-4386-83a8-5b74c3a9a28c" />
just see pic
### Steps to reproduce
1. click on usage
### Expected behavior
nothing is doubled
### Actual behavior
h... | open | null | false | 3 | [
"bug",
"regression"
] | [] | 2026-03-14T03:48:04Z | 2026-05-04T13:19:38Z | null | CONTRIBUTOR | null | 20260504T180039Z | 2026-05-04T18:00:39Z | HenryLoenwind | 1,485,873 | MDQ6VXNlcjE0ODU4NzM= | User | false |
openclaw/openclaw | 4,074,572,466 | I_kwDOQb6kR87y3Qqy | 45,711 | https://github.com/openclaw/openclaw/issues/45711 | https://api.github.com/repos/openclaw/openclaw/issues/45711 | [Bug]: Sess v. now vr. obsc. (UI v2) | ### Bug type
Regression (worked before, now fails)
### Summary
The session view is now very obscure and the displayed fields are shortened to make no sense anymore without any reason.
old:
<img width="1093" height="264" alt="Image" src="https://github.com/user-attachments/assets/3f960d66-306e-4d21-8d99-b0ca18ae74f... | open | null | false | 4 | [
"bug",
"regression"
] | [
"BunsDev"
] | 2026-03-14T03:55:18Z | 2026-05-04T13:19:40Z | null | CONTRIBUTOR | null | 20260504T180039Z | 2026-05-04T18:00:39Z | HenryLoenwind | 1,485,873 | MDQ6VXNlcjE0ODU4NzM= | User | false |
openclaw/openclaw | 4,074,594,704 | I_kwDOQb6kR87y3WGQ | 45,718 | https://github.com/openclaw/openclaw/issues/45718 | https://api.github.com/repos/openclaw/openclaw/issues/45718 | Session bloat: skillsSnapshot and systemPromptReport accumulated on every run | ## Bug Description
Sessions accumulate skillsSnapshot and systemPromptReport fields on every agent run, causing sessions.json to grow unbounded until context overflow errors occur.
## Root Cause
Two issues in the session persistence logic:
1. **systemPromptReport added unconditionally** (agent-Cp6s4ne-.js line ~293... | open | null | false | 3 | [] | [] | 2026-03-14T04:04:05Z | 2026-05-04T13:19:42Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | Jragyn | 7,897,461 | MDQ6VXNlcjc4OTc0NjE= | User | false |
openclaw/openclaw | 4,347,567,703 | I_kwDOQb6kR88AAAABAyKeVw | 74,021 | https://github.com/openclaw/openclaw/issues/74021 | https://api.github.com/repos/openclaw/openclaw/issues/74021 | Support reasoning-field outputs and visible final-answer handling for native reasoning models | # Support reasoning-field outputs and visible final-answer handling for native reasoning models
## Summary
Native reasoning models can return the useful answer in `message.reasoning` while `message.content` is null or incomplete. This makes benchmark scores and OpenClaw compatibility misleading unless the harness and... | open | null | false | 2 | [] | [] | 2026-04-29T03:03:09Z | 2026-05-03T03:46:34Z | null | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | jmystaki-create | 267,217,950 | U_kgDOD-1sHg | User | false |
openclaw/openclaw | 4,347,577,573 | I_kwDOQb6kR88AAAABAyLE5Q | 74,025 | https://github.com/openclaw/openclaw/issues/74025 | https://api.github.com/repos/openclaw/openclaw/issues/74025 | [Feature]: Explicit command for executing taskflows/lobster manually (slash command) | ### Summary
Create a slash command that allows the manual execution of a Task Flow or lobster workflow.
### Problem to solve
Currently there's no way to explicitly instruct an agent to perform a task flow or a lobster workflow. It would be more effective (faster, tokens, etc.) to run these manual in some cases. Espe... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-04-29T03:05:23Z | 2026-05-03T10:19:50Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | technicallysecure | 92,497,344 | U_kgDOBYNlwA | User | false |
openclaw/openclaw | 4,348,024,807 | I_kwDOQb6kR88AAAABAymX5w | 74,086 | https://github.com/openclaw/openclaw/issues/74086 | https://api.github.com/repos/openclaw/openclaw/issues/74086 | [Bug]: Regression: Telegram provider fails on Windows after 2026.4.23 (deleteWebhook / setMyCommands errors) | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After upgrading OpenClaw from 2026.4.23 to any newer version (2026.4.24+), the Telegram provider fails to initialise correctly on Windows. The gateway starts and reaches ready, but Telegram repeatedly fails webhook cleanup a... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-04-29T05:03:03Z | 2026-05-02T04:25:41Z | 2026-04-29T12:48:55Z | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | trentwolfchester | 249,177,450 | U_kgDODtolag | User | false |
openclaw/openclaw | 4,348,060,786 | I_kwDOQb6kR88AAAABAyokcg | 74,087 | https://github.com/openclaw/openclaw/issues/74087 | https://api.github.com/repos/openclaw/openclaw/issues/74087 | Cold-start optimization: ship pre-warmed Node compile cache + provide opt-in to skip self-respawn | ## Summary
OpenClaw's coldstart on stateless container runtimes (Cloud Run, Lambda, Fly Machines, fresh K8s pods) currently spends *~50s in V8 parse+compile* of the bundled chunk graph before any log line is emitted, plus *~5–10s* in a self-respawn fork. Two small upstream changes would let every install benefit:
1. ... | open | null | false | 1 | [] | [] | 2026-04-29T05:10:48Z | 2026-05-03T13:22:06Z | null | NONE | null | 20260503T150031Z | 2026-05-03T15:00:31Z | jarvis-ai-gregmoser | 258,473,251 | U_kgDOD2f9Iw | User | false |
openclaw/openclaw | 4,067,723,438 | I_kwDOQb6kR87ydIiu | 44,424 | https://github.com/openclaw/openclaw/issues/44424 | https://api.github.com/repos/openclaw/openclaw/issues/44424 | Bug: Telegram messages not routed to correct forum topic | ## Description
When using message(action=send) to send a message to a different forum topic (threadId) within the same Telegram group, the request intermittently fails with:
```
Network request for 'sendMessage' failed!
```
Sending messages to the current topic (where the agent session is bound) works fine.
## Step... | closed | not_planned | true | 2 | [] | [] | 2026-03-12T22:48:49Z | 2026-05-02T04:29:23Z | 2026-04-29T05:15:41Z | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | mikewilson9 | 15,689,235 | MDQ6VXNlcjE1Njg5MjM1 | User | false |
openclaw/openclaw | 4,348,110,199 | I_kwDOQb6kR88AAAABAyrldw | 74,096 | https://github.com/openclaw/openclaw/issues/74096 | https://api.github.com/repos/openclaw/openclaw/issues/74096 | image/video/music generate tools eagerly list capability providers, causing ~15-18s reload per turn | ### Bug type
Behavior bug (incorrect output/state without crash) — performance.
### Beta release blocker
No
### Summary
The `image-generate`, `video-generate`, and `music-generate` agent tools eagerly list runtime capability providers during tool registration, even when the agent has an explicit and complete `modelCo... | open | null | false | 1 | [] | [] | 2026-04-29T05:20:57Z | 2026-05-02T22:59:17Z | null | NONE | null | 20260503T000641Z | 2026-05-03T00:06:41Z | poolside-ventures | 215,486,291 | U_kgDODNgPUw | User | false |
openclaw/openclaw | 4,348,109,170 | I_kwDOQb6kR88AAAABAyrhcg | 74,094 | https://github.com/openclaw/openclaw/issues/74094 | https://api.github.com/repos/openclaw/openclaw/issues/74094 | feat(tasks): add compressed operational summaries | ## Summary
Add a compact, actionable summary layer for task status across the task system, CLI, and auto-reply surfaces.
The goal is to make long-running or blocked work understandable in one glance, without requiring users or downstream consumers to reconstruct state from logs or scattered fields.
## Problem
Task ... | open | null | false | 1 | [] | [] | 2026-04-29T05:20:42Z | 2026-05-03T10:19:52Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | neofdezf-cloud | 271,599,963 | U_kgDOEDBJWw | User | false |
openclaw/openclaw | 4,348,110,050 | I_kwDOQb6kR88AAAABAyrk4g | 74,095 | https://github.com/openclaw/openclaw/issues/74095 | https://api.github.com/repos/openclaw/openclaw/issues/74095 | feat(tasks): normalize blocker and failure taxonomy | ## Summary
Introduce a shared blocker/failure taxonomy for tasks so OpenClaw can distinguish between waiting, blocked, failed, and recoverable transient failure in a machine-readable way.
## Problem
Some task states and failure conditions are visible today, but the reason behind a stall or failure is not always norm... | open | null | false | 1 | [] | [] | 2026-04-29T05:20:54Z | 2026-05-03T09:13:30Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | neofdezf-cloud | 271,599,963 | U_kgDOEDBJWw | User | false |
openclaw/openclaw | 4,348,081,632 | I_kwDOQb6kR88AAAABAyp14A | 74,090 | https://github.com/openclaw/openclaw/issues/74090 | https://api.github.com/repos/openclaw/openclaw/issues/74090 | [Feature]: surface promote-then-rotate suggestion when session context overflow trips repeatedly | ### Summary
When session-level context overflow trips repeatedly on the same session (`[context-overflow-diag] compactionAttempts`), surface a one-time `promote-then-rotate` suggestion to the operator with a pre-computed dry-run of `memory promote` candidates, plus a one-click confirm to (a) promote the candidates and... | open | null | false | 1 | [] | [] | 2026-04-29T05:14:51Z | 2026-05-03T01:23:08Z | null | NONE | null | 20260503T030030Z | 2026-05-03T03:00:30Z | Jason-Vaughan | 95,194,903 | U_kgDOBayPFw | User | false |
openclaw/openclaw | 4,348,148,867 | I_kwDOQb6kR88AAAABAyt8gw | 74,100 | https://github.com/openclaw/openclaw/issues/74100 | https://api.github.com/repos/openclaw/openclaw/issues/74100 | [Feature Request] Skill Graph - 按需加载依赖链，减少 Token 消耗 | [Feature Request] Skill Graph - 按需加载依赖链，减少 Token 消耗
## 背景
当前 OpenClaw 启动时加载所有技能定义，导致大量 token 消耗。许多任务只需要 1-2 层 skill 调用，但系统会预加载全部。
## 需求：Skill Graph 依赖按需加载
### 方案
构建 skill 之间的依赖图谱，执行时按需展开：
\`\`\`
用户请求
↓
加载 L1 Skill（如 supply-kitting-analysis）
↓
发现依赖 L2 Skills（如 inventory-query, po-check）
↓
按需加载 L2 → 发现依赖 ... | open | null | false | 1 | [] | [] | 2026-04-29T05:30:19Z | 2026-05-03T08:47:54Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | zhonghuabaima | 1,112,261 | MDQ6VXNlcjExMTIyNjE= | User | false |
openclaw/openclaw | 4,337,538,993 | I_kwDOQb6kR88AAAABAomXsQ | 72,948 | https://github.com/openclaw/openclaw/issues/72948 | https://api.github.com/repos/openclaw/openclaw/issues/72948 | openclaw gateway stop returns 'Gateway service disabled' without killing a foreground-launched gateway | ## Summary
> **In plain English:** if you start the gateway in foreground (`openclaw gateway`, not `openclaw daemon start`), then later run `openclaw gateway stop`, the stop command returns "Gateway service disabled" — but doesn't actually kill the running gateway. The process keeps running and port 18789 stays bound.... | open | null | false | 4 | [] | [] | 2026-04-27T17:10:32Z | 2026-05-03T06:07:37Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | mayank6136 | 16,468,779 | MDQ6VXNlcjE2NDY4Nzc5 | User | false |
openclaw/openclaw | 4,074,704,668 | I_kwDOQb6kR87y3w8c | 45,755 | https://github.com/openclaw/openclaw/issues/45755 | https://api.github.com/repos/openclaw/openclaw/issues/45755 | [Bug]: Agent emoji as AWOL (ui v0.2) | ### Bug type
Regression (worked before, now fails)
### Summary
new:
<img width="134" height="93" alt="Image" src="https://github.com/user-attachments/assets/24db96f1-fc99-4779-91e6-5056d6f6d2ec" />
old:
<img width="204" height="98" alt="Image" src="https://github.com/user-attachments/assets/e4c5ab8a-c766-4fa8-82e... | open | null | false | 3 | [
"bug",
"regression"
] | [] | 2026-03-14T05:06:13Z | 2026-05-04T00:13:00Z | null | CONTRIBUTOR | null | 20260504T030031Z | 2026-05-04T03:00:31Z | HenryLoenwind | 1,485,873 | MDQ6VXNlcjE0ODU4NzM= | User | false |
openclaw/openclaw | 4,074,719,991 | I_kwDOQb6kR87y30r3 | 45,759 | https://github.com/openclaw/openclaw/issues/45759 | https://api.github.com/repos/openclaw/openclaw/issues/45759 | Telegram typing keepalive loop lacks circuit breaker, causes gateway crash on network failure | ## Summary
When the Telegram API becomes unreachable (network blip, DNS timeout, etc.), the typing indicator keepalive loop (`createTypingKeepaliveLoop` in `src/channels/typing-lifecycle.ts`) continues firing `sendChatAction` calls every 6 seconds indefinitely. Each failed call triggers up to 3 retries with exponentia... | open | null | false | 11 | [] | [] | 2026-03-14T05:16:41Z | 2026-05-04T13:19:43Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | assistantheinrich-prog | 261,204,934 | U_kgDOD5Grxg | User | false |
openclaw/openclaw | 4,074,749,297 | I_kwDOQb6kR87y371x | 45,765 | https://github.com/openclaw/openclaw/issues/45765 | https://api.github.com/repos/openclaw/openclaw/issues/45765 | [Bug] OPENCLAW_HOME 设置为 ~/.openclaw 时产生嵌套目录 ~/.openclaw/.openclaw | ### Bug type
Regression (worked before, now fails)
### Summary
当 OPENCLAW_HOME 环境变量被设置为 ~/.openclaw 时，运行 openclaw onboard 会产生嵌套目录 ~/.openclaw/.openclaw，导致配置文件和状态数据被写入错误位置。
### Steps to reproduce
1. 设置环境变量：export OPENCLAW_HOME=/home/tpam/.openclaw
2. 运行 onboarding：openclaw onboard
3. 选择本地 Gateway 配置
4. 完成配置后检查目录结构
... | open | null | false | 7 | [
"bug",
"regression"
] | [] | 2026-03-14T05:29:50Z | 2026-05-04T13:19:45Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | max2055 | 26,732,746 | MDQ6VXNlcjI2NzMyNzQ2 | User | false |
openclaw/openclaw | 4,074,773,027 | I_kwDOQb6kR87y4Boj | 45,771 | https://github.com/openclaw/openclaw/issues/45771 | https://api.github.com/repos/openclaw/openclaw/issues/45771 | Feature: Built-in pace-aware rate limiting for autonomous agents | # Feature Proposal: Built-in Pace-Aware Rate Limiting
## Problem
OpenClaw agents running autonomous loops (subagents, heartbeat-driven coding sessions) can easily burn through Anthropic API rate limits without awareness of their consumption pace. There's no built-in mechanism to:
1. Track current usage against the 5... | open | null | false | 4 | [] | [] | 2026-03-14T05:43:19Z | 2026-05-04T13:19:47Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | o-shabashov | 1,227,282 | MDQ6VXNlcjEyMjcyODI= | User | false |
openclaw/openclaw | 4,074,799,266 | I_kwDOQb6kR87y4ICi | 45,778 | https://github.com/openclaw/openclaw/issues/45778 | https://api.github.com/repos/openclaw/openclaw/issues/45778 | Skills cannot push background messages via crontab — need env injection or directExec cron mode | ## Problem
Skills that use system `crontab` for zero-LLM-cost background monitoring cannot send messages back to the user, because crontab jobs run outside the agent context and don't know the user's chat route (channel type + target ID).
The only working background push mechanism today is `openclaw cron` with `agent... | open | null | false | 6 | [] | [] | 2026-03-14T05:55:15Z | 2026-05-04T13:19:48Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | vahnxu | 19,759,725 | MDQ6VXNlcjE5NzU5NzI1 | User | false |
openclaw/openclaw | 4,074,847,981 | I_kwDOQb6kR87y4T7t | 45,799 | https://github.com/openclaw/openclaw/issues/45799 | https://api.github.com/repos/openclaw/openclaw/issues/45799 | [Bug]: Telegram media download fails silently when proxy unavailable (no retry) | ### Bug type
Regression (worked before, now fails)
### Summary
Telegram inbound media (images/photos sent by users) fails to download when the configured HTTP proxy (127.0.0.1:7890 via ClashX) drops momentarily. No retry is attempted. User sees: ⚠️ Failed to download media. Please try again.
### Steps to reproduce
... | open | null | false | 6 | [
"bug",
"regression"
] | [] | 2026-03-14T06:23:40Z | 2026-05-04T00:13:03Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | rexliu1912-star | 247,500,270 | U_kgDODsCN7g | User | false |
openclaw/openclaw | 4,075,003,663 | I_kwDOQb6kR87y458P | 45,846 | https://github.com/openclaw/openclaw/issues/45846 | https://api.github.com/repos/openclaw/openclaw/issues/45846 | [Bug]: Web UI chat images display too large | ### Bug type
Regression (worked before, now fails)
### Summary
Images in webchat become abnormally large after chatting for a while
### Steps to reproduce
1. Open OpenClaw Web UI
2. Have a conversation in webchat
3. Send/receive multiple messages with images
4. After chatting for a while, images in the chat become... | open | null | false | 3 | [
"bug",
"regression"
] | [] | 2026-03-14T07:31:13Z | 2026-05-04T13:19:50Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | zelixag | 9,192,601 | MDQ6VXNlcjkxOTI2MDE= | User | false |
openclaw/openclaw | 4,075,269,383 | I_kwDOQb6kR87y560H | 45,926 | https://github.com/openclaw/openclaw/issues/45926 | https://api.github.com/repos/openclaw/openclaw/issues/45926 | sessions_send: announce step skipped on timeout (should run async) | ## Summary
When `sessions_send` is called with `timeoutSeconds > 0` and `agent.wait` times out, the announce step is **completely skipped**. The code path returns immediately on timeout without calling `startA2AFlow`, so the target agent never gets the "Agent-to-agent announce step." message and therefore never delive... | closed | completed | false | 2 | [] | [] | 2026-03-14T09:20:03Z | 2026-05-04T13:46:49Z | 2026-05-04T13:46:49Z | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | Chaixueyuan | 30,581,479 | MDQ6VXNlcjMwNTgxNDc5 | User | false |
openclaw/openclaw | 4,075,349,553 | I_kwDOQb6kR87y6OYx | 45,952 | https://github.com/openclaw/openclaw/issues/45952 | https://api.github.com/repos/openclaw/openclaw/issues/45952 | Webchat: messages lost during WebSocket reconnect (no client-side queue/ACK) | ## Problem
Messages sent via the Webchat UI are silently lost when the WebSocket connection drops and reconnects (code 1001 / "Going Away"). This happens when:
- The browser tab reloads or navigates
- The WebSocket briefly disconnects for any reason
- The user hits Enter at the exact moment of a reconnect cycle
The ... | open | null | false | 3 | [] | [] | 2026-03-14T09:51:55Z | 2026-05-04T00:43:33Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | Urberliner | 265,644,424 | U_kgDOD9VpiA | User | false |
openclaw/openclaw | 4,075,450,412 | I_kwDOQb6kR87y6nAs | 45,993 | https://github.com/openclaw/openclaw/issues/45993 | https://api.github.com/repos/openclaw/openclaw/issues/45993 | Feature Request: Cross-platform migration tool (migrate export/import) | ## Summary
Add `openclaw migrate export` and `openclaw migrate import` commands for seamless cross-platform workspace migration (Linux ↔ Windows ↔ macOS).
## Problem
Currently, migrating OpenClaw between machines requires:
1. Manually copying `~/.openclaw/workspace/` files (SOUL.md, AGENTS.md, skills/, etc.)
2. Manu... | open | null | false | 2 | [] | [] | 2026-03-14T10:34:42Z | 2026-05-04T13:19:59Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | cui-yan | 25,812,973 | MDQ6VXNlcjI1ODEyOTcz | User | false |
openclaw/openclaw | 4,075,503,492 | I_kwDOQb6kR87y6z-E | 46,015 | https://github.com/openclaw/openclaw/issues/46015 | https://api.github.com/repos/openclaw/openclaw/issues/46015 | [Bug]: messages.tts ElevenLabs provider silently falls back to Edge TTS in Discord; no per-agent voice support | ### Bug type
Regression (worked before, now fails)
### Summary
messages.tts.provider: elevenlabs configured correctly, API key valid, voiceId accessible
/voice audio works correctly with ElevenLabs
But auto TTS always produces Edge/Microsoft voice, no error in logs
Telegram supports per-agent voice assignment; Disco... | open | null | false | 3 | [
"bug",
"regression"
] | [] | 2026-03-14T10:56:05Z | 2026-05-04T01:19:49Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | alainlx2024 | 163,292,621 | U_kgDOCbulzQ | User | false |
openclaw/openclaw | 4,075,523,132 | I_kwDOQb6kR87y64w8 | 46,023 | https://github.com/openclaw/openclaw/issues/46023 | https://api.github.com/repos/openclaw/openclaw/issues/46023 | [Bug]: Telegram final replies with media ignore configured mediaMaxMb in outbound delivery path | ### Bug type
Crash (process/app exits or hangs)
### Summary
Telegram final replies that include media do not honor the configured mediaMaxMb because the outbound delivery path does not propagate `mediaMaxBytes` into `deliverMediaReply(...)` /` loadWebMedia(...)`.
### Steps to reproduce
1. Install OpenClaw via glob... | open | null | false | 3 | [
"bug",
"bug:crash"
] | [] | 2026-03-14T11:04:26Z | 2026-05-04T13:20:01Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | Designdocs | 54,241,667 | MDQ6VXNlcjU0MjQxNjY3 | User | false |
openclaw/openclaw | 4,075,537,750 | I_kwDOQb6kR87y68VW | 46,031 | https://github.com/openclaw/openclaw/issues/46031 | https://api.github.com/repos/openclaw/openclaw/issues/46031 | auth.order ignored for GitHub Copilot provider — first profile in auth.profiles always wins | ## Bug Report
### Summary
`auth.order` configuration is ignored when selecting the active GitHub Copilot auth profile. The runtime always uses the **first key** in `auth.profiles` object (JSON insertion order), regardless of what is specified in `auth.order`.
### Environment
- OpenClaw version: 2026.3.1
- Platform:... | open | null | false | 4 | [] | [] | 2026-03-14T11:11:07Z | 2026-05-04T13:20:03Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | AtaraxiaEta | 11,269,851 | MDQ6VXNlcjExMjY5ODUx | User | false |
openclaw/openclaw | 4,075,638,786 | I_kwDOQb6kR87y7VAC | 46,056 | https://github.com/openclaw/openclaw/issues/46056 | https://api.github.com/repos/openclaw/openclaw/issues/46056 | Shell builtins (e.g. cd) always trigger approval gate even when allowlist is configured | ## Summary
Shell builtins such as `cd`, `export`, `echo` (when used as builtins) always trigger the exec approval gate, even when all other binaries in the command chain are correctly allowlisted. This behaviour is undocumented.
## Reproduction
With `/usr/bin/git` and `/usr/bin/bash` added to the agent allowlist:
`... | open | null | false | 3 | [] | [] | 2026-03-14T11:48:08Z | 2026-05-04T13:20:05Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | aukei | 26,145,869 | MDQ6VXNlcjI2MTQ1ODY5 | User | false |
openclaw/openclaw | 4,075,642,159 | I_kwDOQb6kR87y7V0v | 46,058 | https://github.com/openclaw/openclaw/issues/46058 | https://api.github.com/repos/openclaw/openclaw/issues/46058 | Exploring a chat-first Android surface for OpenClaw (discussion + possible focused upstreaming) | Hi OpenClaw team,
I’ve been building an independent Android fork from `openclaw/openclaw` (`apps/android`) to validate a narrower use case: a direct, chat-first mobile surface for interacting with OpenClaw agents.
I’m **not** asking to upstream the fork as a whole.
I’d like to ask whether maintainers would be open ... | open | null | false | 4 | [] | [] | 2026-03-14T11:50:03Z | 2026-05-04T01:32:04Z | null | CONTRIBUTOR | null | 20260504T030031Z | 2026-05-04T03:00:31Z | memphislee09-source | 261,085,282 | U_kgDOD4_YYg | User | false |
openclaw/openclaw | 4,075,666,776 | I_kwDOQb6kR87y7b1Y | 46,080 | https://github.com/openclaw/openclaw/issues/46080 | https://api.github.com/repos/openclaw/openclaw/issues/46080 | Bug: Anthropic tool_result succeeds but final assistant content is empty, causing 'No reply from agent' | ## Summary
When using an Anthropic-compatible custom provider via `anthropic-messages`, OpenClaw successfully completes the tool phase (`tool_use` emitted, tool executed, `toolResult` recorded), but the final assistant turn is persisted with `content: []` and `stopReason: "stop"`. The user sees `No reply from agent.` ... | open | null | false | 4 | [
"bug",
"agents"
] | [
"vincentkoc"
] | 2026-03-14T12:03:52Z | 2026-05-04T13:20:06Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | wmgx | 38,006,179 | MDQ6VXNlcjM4MDA2MTc5 | User | false |
openclaw/openclaw | 4,075,680,191 | I_kwDOQb6kR87y7fG_ | 46,086 | https://github.com/openclaw/openclaw/issues/46086 | https://api.github.com/repos/openclaw/openclaw/issues/46086 | memorySearch.multimodal requires provider="gemini" but worked with local in 2026.2.12 | ## Description
After upgrading from 2026.2.12 to 2026.2.13 (current: 2026.3.13), memorySearch with multimodal enabled no longer works with local provider.
| open | null | false | 2 | [] | [] | 2026-03-14T12:10:20Z | 2026-05-04T13:20:08Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | alexurui | 130,198,435 | U_kgDOB8Krow | User | false |
openclaw/openclaw | 4,075,715,270 | I_kwDOQb6kR87y7nrG | 46,101 | https://github.com/openclaw/openclaw/issues/46101 | https://api.github.com/repos/openclaw/openclaw/issues/46101 | doctor: warn when agents.defaults.heartbeat is shadowed by per-agent blocks | ## Problem
When any agent in `agents.list[]` has an explicit `heartbeat` block, OpenClaw only runs heartbeats for agents with explicit blocks. Agents relying on `agents.defaults.heartbeat` are silently excluded — no warning in logs, no `openclaw doctor` finding, no runtime indication.
This is a footgun in multi-agent... | open | null | false | 2 | [] | [] | 2026-03-14T12:27:46Z | 2026-05-04T00:43:37Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | sene1337 | 261,265,124 | U_kgDOD5KW5A | User | false |
openclaw/openclaw | 4,075,734,140 | I_kwDOQb6kR87y7sR8 | 46,109 | https://github.com/openclaw/openclaw/issues/46109 | https://api.github.com/repos/openclaw/openclaw/issues/46109 | Feature request: first-class .env / environment-backed secret config UX | # Feature request: first-class `.env` / environment-backed secret config UX
## Summary
OpenClaw already supports environment-backed secrets well enough for runtime, but the UX still nudges users toward storing secrets directly in `openclaw.json`.
Using `.env` / environment variables is materially better for:
- secu... | open | null | false | 3 | [] | [] | 2026-03-14T12:34:53Z | 2026-05-04T02:38:13Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | SakenW | 16,665,726 | MDQ6VXNlcjE2NjY1NzI2 | User | false |
openclaw/openclaw | 4,075,818,226 | I_kwDOQb6kR87y8Azy | 46,153 | https://github.com/openclaw/openclaw/issues/46153 | https://api.github.com/repos/openclaw/openclaw/issues/46153 | Gateway hangs silently on startup when launched by launchd (macOS) | ## Summary
When the gateway is started by macOS launchd (via LaunchAgent), the Node.js process loads but **hangs forever during module evaluation** — never binding a port, producing any output, or responding. The exact same command, environment, and arguments work correctly when run manually (including via `nohup`, `e... | open | null | false | 2 | [] | [] | 2026-03-14T13:23:06Z | 2026-05-04T01:19:54Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | frankr | 134,367 | MDQ6VXNlcjEzNDM2Nw== | User | false |
openclaw/openclaw | 4,075,965,164 | I_kwDOQb6kR87y8krs | 46,252 | https://github.com/openclaw/openclaw/issues/46252 | https://api.github.com/repos/openclaw/openclaw/issues/46252 | Cost dashboard omits .jsonl.reset.<timestamp> archive files, severely undercounting daily spend for users of /new | ## Summary
The OpenClaw cost dashboard/UI only aggregates **active** `.jsonl` session files when computing usage and cost. When a user runs `/new` (or `/reset`), the current session file is **renamed** to `<uuid>.jsonl.reset.<ISO-timestamp>` — it is not deleted. Those archive files are never included in the dashboard ... | open | null | false | 5 | [] | [] | 2026-03-14T14:51:04Z | 2026-05-04T01:32:05Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | PabloDaVa | 145,587,051 | U_kgDOCK17aw | User | false |
openclaw/openclaw | 4,076,178,960 | I_kwDOQb6kR87y9Y4Q | 46,341 | https://github.com/openclaw/openclaw/issues/46341 | https://api.github.com/repos/openclaw/openclaw/issues/46341 | [Bug] Recurring cron jobs silently skip transient-error retry — only one-shot jobs retry with backoff | ### Summary
One-shot (`at`) jobs correctly detect transient errors (network, timeout, rate-limit, 5xx) and retry with exponential backoff. Recurring (`cron` / `every`) jobs have no such logic — on any error they simply wait until the next natural schedule time, which can be hours or days away.
### Root Cause
In `src... | open | null | false | 3 | [] | [] | 2026-03-14T16:27:44Z | 2026-05-04T13:20:10Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | onesmash | 3,243,910 | MDQ6VXNlcjMyNDM5MTA= | User | false |
openclaw/openclaw | 4,076,208,773 | I_kwDOQb6kR87y9gKF | 46,369 | https://github.com/openclaw/openclaw/issues/46369 | https://api.github.com/repos/openclaw/openclaw/issues/46369 | [Bug]: cron.remove does not clean up session records in sessions.json | ### Bug type
Regression (worked before, now fails)
### Summary
When deleting a cron job via cron.remove, the session record remains in sessions.json, causing deleted jobs to still appear in the Control UI session dropdown.
### Steps to reproduce
1. Create a cron job via Control UI or cron.add
2. Delete the cron jo... | open | null | false | 2 | [
"bug",
"regression"
] | [] | 2026-03-14T16:41:04Z | 2026-05-04T01:32:08Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | 15881266296-commits | 243,777,707 | U_kgDODofAqw | User | false |
openclaw/openclaw | 4,076,216,122 | I_kwDOQb6kR87y9h86 | 46,378 | https://github.com/openclaw/openclaw/issues/46378 | https://api.github.com/repos/openclaw/openclaw/issues/46378 | [Bug]: 安装配置界面windows会卡死 | ### Bug type
Regression (worked before, now fails)
### Summary
🦞 OpenClaw 2026.3.13 (61d171a) — Your .env is showing; don't worry, I'll pretend I didn't see it.
▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄
██░▄▄▄░██░▄▄░██░▄▄▄██░▀██░██░▄▄▀██░████░▄▄▀██░███░██
██░███░██░▀▀░██░▄▄▄██░█░█░██░█████░████░▀▀░██░█░... | open | null | false | 2 | [
"bug",
"regression"
] | [] | 2026-03-14T16:44:48Z | 2026-05-04T13:20:12Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | sdadasaf | 39,111,052 | MDQ6VXNlcjM5MTExMDUy | User | false |
openclaw/openclaw | 4,076,416,695 | I_kwDOQb6kR87y-S63 | 46,494 | https://github.com/openclaw/openclaw/issues/46494 | https://api.github.com/repos/openclaw/openclaw/issues/46494 | [Bug]: `openclaw health --json` reports Telegram `running: false` / `tokenSource: "none"` while `channels status --json` reports `running: true` / `tokenSource: "config"` | ## Bug type
Behavior bug
## Summary
`openclaw health --json` reports a configured Telegram channel as not running with `tokenSource: "none"` even though `openclaw channels status --json`, `openclaw channels status --probe`, and live Telegram traffic all show the same account running normally from config.
## Steps t... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-14T18:22:46Z | 2026-05-02T04:27:08Z | 2026-04-29T21:26:32Z | CONTRIBUTOR | null | 20260502T060142Z | 2026-05-02T06:01:42Z | 0xble | 20,782,088 | MDQ6VXNlcjIwNzgyMDg4 | User | false |
openclaw/openclaw | 4,076,481,082 | I_kwDOQb6kR87y-io6 | 46,518 | https://github.com/openclaw/openclaw/issues/46518 | https://api.github.com/repos/openclaw/openclaw/issues/46518 | WhatsApp QR pairing fails silently on 2026.3.13 (macOS ARM64 Docker) | ## Environment
- **OpenClaw:** 2026.3.13 (official `ghcr.io/openclaw/openclaw:latest`, commit 70d7a08)
- **Also tested:** 2026.3.8, 2026.3.11, 2026.3.12 (coollabsio images) — same behavior
- **Platform:** macOS Sequoia 25.3.0, Apple M3 Max (ARM64)
- **Docker:** Docker Desktop for Mac
- **Phone:** iPhone, latest WhatsA... | open | null | false | 3 | [] | [] | 2026-03-14T18:56:09Z | 2026-05-04T01:32:09Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | innersanctumtech | 149,255,110 | U_kgDOCOVzxg | User | false |
openclaw/openclaw | 4,076,550,684 | I_kwDOQb6kR87y-zoc | 46,531 | https://github.com/openclaw/openclaw/issues/46531 | https://api.github.com/repos/openclaw/openclaw/issues/46531 | [Bug]: model-assisted Moonshot/Kimi setup can write a daemon-invalid secret reference and crash-loop the gateway | ### Bug type
Regression (worked before, now fails)
### Summary
Bug type:
Regression (worked before, now fails)
Summary:
Using the in-product assistant/model workflow to add Moonshot/Kimi caused OpenClaw to write a config that referenced ${MOONSHOT_API_KEY}, but the systemd gateway daemon could not see that env var.... | open | null | false | 3 | [
"bug",
"regression"
] | [] | 2026-03-14T19:28:11Z | 2026-05-04T13:20:14Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | alexrmacleod | 7,895,493 | MDQ6VXNlcjc4OTU0OTM= | User | false |
openclaw/openclaw | 4,076,596,096 | I_kwDOQb6kR87y--uA | 46,544 | https://github.com/openclaw/openclaw/issues/46544 | https://api.github.com/repos/openclaw/openclaw/issues/46544 | Feature: CLI 'message send --media' should use agent-scoped media roots instead of default roots | ## Problem
When using \openclaw message send --media <local-path>\ from the CLI, the command uses \getDefaultMediaLocalRoots()\ which only allows:
- \~/.openclaw/media\
- \~/.openclaw/agents\
- \~/.openclaw/workspace\ (only the base workspace, not \workspace-*\ variants)
- \~/.openclaw/sandboxes\
- System temp directo... | closed | completed | true | 3 | [] | [] | 2026-03-14T19:52:49Z | 2026-05-03T04:35:54Z | 2026-04-30T17:03:50Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | alprclbi | 246,855,805 | U_kgDODra4fQ | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.