ME News message, April 23 (UTC+8), according to Beating monitoring, the open-source AI Agent platform OpenClaw released version 2026.4.22. The biggest change is that the Codex harness is aligned with Pi's lifecycle. Previously, plugins behaved inconsistently on the Pi harness and Codex harness paths, with the same plugin potentially missing hook calls under different harnesses. This version fills in key hooks such as before_prompt_build, \before_compaction / after_compaction, \after_tool_call, \before_message_write, \llm_input / llm_output / agent_end, etc., so plugin developers no longer need to adapt separately for the two paths. It also adds a plugin extension interface on the Codex side, supporting asynchronous tool_result middleware. This is another round of systematic patch by OpenClaw on Codex integration, following this week's OpenAI Codex team's fix of the Codex harness authentication silent fallback issue. Another architectural addition is the TUI local embedding mode: users can run agent conversations directly in the terminal without starting Gateway, while retaining the plugin approval mechanism. The default thinking level of reasoning models has been silently upgraded from the previous off/low to medium. Users who have not manually set the reasoning level will find that the reasoning model starts outputting thinking processes by default after upgrading. In terms of performance, plugin loading now uses native Jiti, reducing startup time by 82% to 90%; \doctor --non-interactive\ runtime reduced by about 74%. Kimi K2.6 multi-turn agent calls no longer get interrupted due to incorrect cleansing of tool_call IDs. On Linux, subprocesses automatically increase \oom_score_adj, so under memory pressure the kernel preferentially terminates temporary workers rather than the Gateway main process. The configuration system adds last-known-good recovery to prevent accidental config overwrites from causing Gateway crash loops. On the provider side, xAI adds image generation (grok-imagine-image / grok-imagine-image-pro), TTS and STT; Tencent Cloud is built-in as an official provider plugin, with Hy3 preview model and pricing included. When the OpenAI model enables web search, it now directly uses OpenAI's native web_search tool, no longer going through OpenClaw's hosted search channel. (Source: BlockBeats)

View Original

This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.

Reward
like
Comment
Repost
Share

Comment

Add a comment

No comments

Trending Topics
View More
#
Get2SharesOfSKHynixAtZeroCost
1.65M Popularity
#
MicronOvertakesMetaInMarketValue
355.71K Popularity
#
WorldCup🇿🇦vs🇨🇦
129.34K Popularity
#
USMayPCEInflationRisesTo4.1%HighestIn3Years
604.09K Popularity
#
StakeUSD1Earn9.48%APR
1M Popularity

Pinned

Sitemap

OpenClaw fixes plugin ecosystem fragmentation: Codex and Pi hooks unified, loading time reduced by 90%

Trending Topics

Get2SharesOfSKHynixAtZeroCost

MicronOvertakesMetaInMarketValue

WorldCup🇿🇦vs🇨🇦

USMayPCEInflationRisesTo4.1%HighestIn3Years

StakeUSD1Earn9.48%APR

Pinned