Claude Code 2.1.139 has been released. 50 CLI changes, 1 system prompt change Highlights: • Added /goal command to run tasks across turns until a set completion condition, with live elapsed/turns/tokens • System prompt compaction is now silent (no trimming alert), which may obscure that prompts were trimmed • Compaction prompt asks model to preserve sensitive instructions, reducing loss of user intent during trimming Full details are in thread ↓

Claude Code CLI 2.1.139 changelog: New features: • Added agent view (Research Preview): a single list of every Claude Code session — running, blocked on you, or done. Run claude agents to get started. See code.claude.com/docs/en/agent-… • Added /goal command: set a completion condition and Claude keeps working across turns until it's met. Works in interactive, -p, and Remote Control. Shows live elapsed/turns/tokens as an overlay panel • Added /scroll-speed command to tune mouse wheel scroll speed with a live preview • Added claude plugin details to show a plugin's component inventory and projected per-session token cost • Added transcript view navigation: ? for keyboard shortcuts, {/} to jump between user prompts, v to toggle shortcut panel • Added hook args: string[] field (exec form) that spawns the command directly without a shell, so path placeholders never need quoting • Added hook continueOnBlock config option for PostToolUse — set to true to feed the hook's rejection reason back to Claude and continue the turn Fixes: • Fixed a deadlock where expired credentials and the forceRemoteSettingsRefresh policy setting blocked claude auth login/logout/status with no way to recover • Fixed autoAllowBashIfSandboxed not auto-approving commands with shell expansions like ＄VAR and $(cmd) • Fixed a bug where a hook writing to the terminal could corrupt an on-screen interactive prompt; hooks now run without terminal access • Fixed unbounded memory growth when an HTTP/SSE MCP server streams non-protocol data — response bodies now capped at 16 MB per SSE frame • Fixed Skill(name *) permission rules — the wildcard form now works as a prefix match, matching Bash(ls *) behavior • Fixed settings hot-reload not detecting edits to symlinked ~/.claude/settings.json • Fixed plugin details failing to load when the marketplace key differs from the manifest name • Fixed /model picker "Default" row not reflecting ANTHROPIC_DEFAULT_OPUS_MODEL/ANTHROPIC_DEFAULT_SONNET_MODEL overrides • Fixed spurious "stream idle timeout" 5 minutes after a response completed, caused by the watchdog timer not being cleared on stream cancellation • Fixed silent exit 1 when 10+ MCP servers are configured and the cache directory is unwritable — the error message now includes the underlying cause • Fixed a typing cursor blinking on tab names, list pointers, and select rows in dialogs • Fixed transcript view letter shortcuts not working after mouse click • Fixed Bash-mode up-arrow history repeating the first entry and clobbering the in-progress draft • Fixed pasting or dropping multiple images only inserting the last one • Fixed hyperlinks using unreadable dark navy on dark themes — they now adapt to the active theme • Fixed model picker showing a redundant "Current model" row for third-party users whose model is set to the opus alias • Fixed legacy Opus picker entry on PAYG 3P providers resolving to the same model as the default entry • Fixed mouse wheel scrolling speed in Cursor and VS Code 1.92–1.104; the trackpad now scrolls at a steady rate and the mouse wheel keeps ~3 lines per notch • Fixed scroll behavior in Windows Terminal and VS Code when attached to background sessions • Fixed MCP resources from disconnected servers lingering in @server: autocomplete • Fixed two-file diff snippets over-reporting the number of truncated lines by one • Fixed Grep results not relativizing Windows drive-letter paths and count mode reporting wrong totals for single-file paths • Fixed border-embedded text overflowing on CJK/emoji due to visual cell width miscalculation • Fixed fuzzy-match highlighting splitting emoji and astral-plane characters mid-pair • Fixed skill argument names containing regex metacharacters breaking argument substitution • Fixed ProgressBar rendering a full block for an almost-full fractional cell • Fixed task polling and fs.watch being resurrected when the last subscriber leaves while a fetch is in flight • Fixed plugin dependency resolution leaving a stale count when the manifest name differs from the source identifier • Fixed Insights Time-of-Day chart skewing when a session has an unparseable timestamp • Fixed keybindings using only the cmd/super/win modifier being flagged as unparseable • Fixed claude_code.active_time.total OpenTelemetry metric not being emitted in --print mode • Fixed claude plugin update not preserving cross-plugin symlinks inside a marketplace Security/safety change: • Remote Control, /schedule, claude.ai MCP connectors, and notification preferences are now disabled when ANTHROPIC_API_KEY / apiKeyHelper / ANTHROPIC_AUTH_TOKEN is set, even if a Claude.ai login also exists. Unset the API key to use these features Other changes: • MCP stdio servers now receive CLAUDE_PROJECT_DIR in their environment, matching hooks. Plugin configs can reference ＄{CLAUDE_PROJECT_DIR} in commands • Compaction prompt now asks the model to preserve sensitive user instructions • /mcp Reconnect now picks up .mcp.json edits without a restart, and shows the HTTP status and URL when reconnecting fails • /context all per-skill token estimates now account for the model's tokenizer and show rounded values • claude plugin install @ now auto-refreshes the marketplace and retries before reporting a plugin as not found • /plugin installed-plugin details now show hook event names and MCP server names cleanly • /context now shows the providing plugin's name for plugin-sourced skills • Remote MCP server reconnect retry on transient failures is now enabled for all users • API requests from subagents now carry x-claude-code-agent-id / x-claude-code-parent-agent-id headers, and claude_code.llm_request OTEL spans include agent_id / parent_agent_id attributes • [VSCode] Press Cmd/Ctrl+Shift+T to reopen the most recently closed session tab, configurable via claudeCode.enableReopenClosedSessionShortcut Source: #21139" target="_blank" rel="nofollow noopener">github.com/anthropics/cla…

Shop kitchen essentials at Amazon 🍳

@ClaudeCodeLog X) goal is awesome, sick and tired of “routines” In desktop app and loop in cli

@ClaudeCodeLog Babysitting multi-turn refactors is exactly what I needed off my plate. Going to point /goal at my migration backlog tonight and see how far it gets before I have to intervene.

@ClaudeCodeLog You know openai codex brought out /goal first to mimic Ralph script from Claude and openly stated so with the release. Why not say the same with /goal you mimicking codex /goal command.

@ClaudeCodeLog Curious if the /goal command is actually useful in practice. I’ve found that launching a fresh CLI instance for each task in the plan, one after another, while tracking progress in the plan itself, works much better for me.

@ClaudeCodeLog /goal is the one. Set a target condition, let it run turns, check the token count when you're back. Cuts the babysitting on multi-step builds.

@ClaudeCodeLog At this point you guys should realy look how to make Claude execute agentic tasks, not lose context, and not act so dumb. Bringing piles of useless features has made the Claude code way worde then 1 yesr ago. That is the reason that many of users swap to codex

Silent compaction is the one to watch very carefully once this is in production. What we have seen: trimmed prompts drop constraints that were quietly holding the agent's behaviour in shape. It keeps running, looks completely normal, and then around turn 10 to 12 it does something it should not. By then the trail is gone. Has anyone mapped how much of the original system prompt intent actually survives a full compaction on a long run?

@ClaudeCodeLog /goal is straight 🔥 Set the finish line and let Claude cook across turns with live time/turns/tokens overlay. No more babysitting. Silent compaction is the perfect stealth win too. Massive upgrade 👏

Memorial Day Sale: Save big + free AI and options data 🗽

@ClaudeCodeLog silent compaction is the one to watch. the trimming alert was the breadcrumb when context drifted mid-session — now you'll just notice something the model knew an hour ago is suddenly gone

@ClaudeCodeLog Spent today running /goal against a legacy .NET monolith. 200 failing tests, one sentence, ~38 seconds. Wrote up the honest take on where it fits and where it doesn't: x.com/jerryharlow/st…

@ClaudeCodeLog @opencode when do you plan to ship /goal?

@ClaudeCodeLog With $200 monthly plan now you can do less than the half of work you could do a month ago, what’s going on??

@ClaudeCodeLog Claude Code is moving fast. That /goal command sounds like it could be really useful for longer tasks.

@ClaudeCodeLog The pace Claude Code is improving right now is honestly insane. The "/goal" command alone feels like a huge shift toward autonomous workflows instead of single prompts

@ClaudeCodeLog Anthropic be like

@ClaudeCodeLog '/goal' is one of those features that sounds small until you realize it turns the CLI from 'assistant per turn' into 'operator with a completion condition.' That's a pretty big workflow shift.

@ClaudeCodeLog they keep adding features while the core logic is still broken. silent prompt trimming? that's just hiding data loss.

@ClaudeCodeLog what's the main use case for the /goal command?

@ClaudeCodeLog Silent compaction is the real story. When agents run on 2000+ line CLAUDE.md files, critical guardrails get trimmed with no signal. Behavior drifts and you have no idea why. "Preserve sensitive instructions" is heuristic. Give me an explicit callback when constraints get dropped.

🤩 Morning can be your favorite part of the day! No matter if you wake up at 5am or 10am, you can make your morning productive and pleasant. Give Fabulous a try! More than 40,000,000+ members and counting! Try the Fabulous app today!

@ClaudeCodeLog Nice one, congrats on the release! The /goal command sounds ace and silent system prompt compaction will cut heaps of noise. Keen to hear if you've noticed any gotchas so far?

@ClaudeCodeLog The /goal command fills a real gap. Previously you had to babysit agent loops manually to catch drift. Is the turn/token ceiling configurable or hard-capped? That determines whether it handles 500-step refactors without manual checkpoints.

@ClaudeCodeLog Let him cook

@ClaudeCodeLog /goal is the one to watch. Loop-until-done at the CLI level means you can wire it into a CI pipeline that retries autonomously. What's the token ceiling before it hard-stops?

@ClaudeCodeLog Everyday is a Claude code day, amazing

@ClaudeCodeLog The /goal until-condition is exactly what my autonomous Claude Code loops needed. I was running Freqtrade backtests in a while-loop with grep on exit codes — the live elapsed/turns/tokens display replaces that whole hack.

@ClaudeCodeLog @grok 总结并发给我。

@ClaudeCodeLog Silent compaction worries me more than the new command excites me. Long agent runs need visible memory boundaries, otherwise you debug behavior without knowing which context vanished.

@ClaudeCodeLog 哇claude-code也有/goal了

@ClaudeCodeLog Claude always copy what other are doing lol

@ClaudeCodeLog didn't you say loop is goal?

@ClaudeCodeLog Finally /goal in cc

Still thinking about getting your kid a bike? 🚲 Summer goes by fast — and there’s no better time for kids to get outside, ride with friends, and build confidence with Guardian Bikes.