Highlights
-
Timed-Out Shell Commands Keep Running: A shell command still running when its timeout expires now moves to the background instead of being killed, so long builds and test runs finish and report back. A foreground
timeout: 0now means the default 120 seconds. (#63699) -
Compaction Without the Wait: As a session nears automatic compaction, the summary is now written in the background, so your next prompt doesn't pause for it. Set
compaction.background: falseto turn it off; a background summary that ends up unused still costs tokens. (#64290) -
GPT 6.2 with ChatGPT Sign-In: GPT 6.2 Sol and GPT 6.2 Astra, plus their fast variants, now show up in the model picker when you sign in with ChatGPT. (#64508)
-
Update Plugins from Settings: Open Settings › Extensions, click Check for updates, then Update on any plugin with a new version. (#64511)
-
Comment on Selected Code: Selecting code in diff and file viewers now just selects it, so you can copy it, and an Add comment toolbar appears beside the selection. (#64410)
Core
Improvements
-
Shell commands still running when their timeout expires now move to the background instead of being killed, in every build. The
shell_timeout_backgroundflag is gone; configs that still set it keep loading. A foregroundtimeout: 0now uses the default 120 second timeout instead of waiting until the command exits. (#63699) -
Background compaction is now on by default: as a session nears automatic compaction, the summary is written in the background so the next prompt doesn't wait for it. Turn it off with
compaction.background: false; a background summary that is never used still costs tokens. Theexperimental.background_compactionsetting is removed. (#64290) -
OpenCode again reads
CLAUDE.mdin any directory that has noAGENTS.md, and falls back to~/.claude/CLAUDE.mdwhen the globalAGENTS.mdis missing. This also applies to instructions found when the agent reads nested files. SetOPENCODE_DISABLE_CLAUDE_CODEorOPENCODE_DISABLE_CLAUDE_CODE_PROMPTto skip everyCLAUDE.md. (#64474, #64533) -
Added GPT 6.2 Sol and GPT 6.2 Astra, plus their fast variants, to the ChatGPT model list. (#64508)
-
websearchnow returns a flat list of results, each with itsurl,title,contentandpublisheddate. (#64471) -
Code Mode and plugins get a simpler structured result from
read:text(withoffset,lineCountandnext),directory(entry names) ormedia(a URI instead of base64). What the model sees is unchanged. (#64361) -
Code Mode programs can use Node's
Bufferfor encoding and decoding bytes. (#64365) -
In Code Mode,
search("issues")now works as shorthand forsearch({ query: "issues" }). (#64347) -
Code Mode tool summaries in the catalog can now be up to 500 characters instead of 120, so the model sees more of each tool's description. (#64341)
-
The
globandgreptools now report a malformed filename glob (for example[) as an error that names the parameter to fix, instead of returning no results. (#63960) -
Server requests that crash now return a 500 with a readable JSON
UnknownErrorbody instead of an empty one. Storage problems name the cause (for example "Storage is full" or "Storage is read-only"); other crashes include arefyou can find in the server log. (#64022) -
The shell tool description now tells the model that its session ID is in the
OPENCODE_SESSION_IDenvironment variable. (#63718) -
Experimental schedules are now automations, behind the
automationsflag (the oldschedulesflag still works). An automation runs on a time, an interval, a cron or a pushed MCP event, and either sends a prompt (to the creating session, another session or a new subagent) or runs a Code Mode program that decides whether to send one. Use/automationor theautomation_*tools. Active or paused plain prompt schedules carry over; schedules with a condition or run limit, and finished ones, are dropped. (#64160)
Bug fixes
-
Fixed finished shell commands leaving zombie processes under
opencode-pty. The process daemon now cleans each command up when it exits. Background processes started by a successful command (cmd &) keep running, and a failed command's leftovers are stopped. (#64266, #64251) -
Fixed ChatGPT and OpenAI subscription logins losing their sharing-limit safeguards and context caps when a token refresh failed. (#64470)
-
Fixed Claude Agent models falling back to a 200K context window when Claude Code adds or renames a model. Thanks @Tarquinen. (#64107)
-
Fixed requests being rejected after switching models mid-session on OpenAI-compatible providers. Models that declare a different reasoning field (such as
reasoning_content) no longer receive