Feature flags, all 2,081.

tengu_groovy_eclipse ◇ Two new internal feature gates were added: tengu_groovy_eclipse and tengu_lucky_quillv2.1.267 → v2.1.308 ◆in the build AND v2.1.267 tengu_grove_oauth_401_received ◇ No description from anybody.never mined ·never mined never read tengu_grove_policy_dismissed ◇ An analytics event, not a feature flag. Claude Code logs it with state: true when the user defers the consumer terms notice during its grace period, either by choosing Not now or by pressing Escape.never mined ·never mined never read tengu_grove_policy_escaped ◇ An analytics event, not a feature flag. Claude Code logs it when the user presses escape in the Consumer Terms and Privacy Policy update dialog (the Grove dialog). It carries no fields.never mined ·never mined never read tengu_grove_policy_exited ◇ An analytics event, not a feature flag. Claude Code logs it during the startup sequence when the Grove policy dialog (location onboarding or policy_update_modal) closes with escape, just before that startup step returns stop. It carries no fields.never mined ·never mined never read tengu_grove_policy_submitted ◇ An analytics event, not a feature flag. Claude Code logs it when the user answers the Grove policy dialog by opting in or out of model training, after saving the choice. It carries state (true for opt-in, false for opt-out) and dismissable (the notice's notice_is_grace_period value).never mined ·never mined never read tengu_grove_policy_toggled ◇ An analytics event, not a feature flag. Claude Code logs it from the privacy settings command when the "Help improve our AI models" setting, re-fetched after the dialog closes, differs from the non-null value it had before. It carries state (the new value) and location settings.never mined ·never mined never read tengu_grove_policy_viewed ◇ An analytics event, not a feature flag. Claude Code logs it when the Grove policy dialog decides to show the Consumer Terms and Privacy Policy update notice. It carries location and dismissable (the notice's notice_is_grace_period value).never mined ·never mined never read tengu_grove_print_viewed ◇ An analytics event, not a feature flag. Claude Code logs it at startup when the Consumer Terms and Privacy Policy update notice is due and is printed as text instead of shown as a dialog. It carries dismissable; when that is not true the printed message starts with [ACTION REQUIRED], otherwise it is a reminder to run claude and review the terms.never mined ·never mined never read tengu_grove_privacy_settings_viewed ◇ An analytics event, not a feature flag. Claude Code logs it when the privacy settings panel titled Data privacy mounts, which the privacy settings command shows when the user's "Help improve our AI models" choice is already set. It carries no fields.never mined ·never mined never read tengu_guest_passes_upsell_shown ◇ An analytics event, not a feature flag. Claude Code logs it when the guest passes upsell line pointing at /passes is shown, after incrementing the stored passesUpsellSeenCount. It carries seen_count, the new count; the upsell is no longer eligible once the count reaches 3 or the user has visited /passes.never mined ·never mined never read tengu_guest_passes_visited ◇ An analytics event, not a feature flag. Claude Code logs it each time the user opens the /passes command. It carries is_first_visit, true on the first visit, which is also when Claude Code stores hasVisitedPasses.never mined ·never mined never read tengu_gypsum_kite ◇ Default false. When on and the account is on the pro plan, the Opus entries in the model picker get · ~2× usage vs Sonnet appended to their description.v2.1.138 or earlier → v2.1.308 ◆in the build AND v2.1.91 tengu_gzip_request_bodies ◇ The gzip decision now records exactly why it skipped, though compression stays off without a server value.v2.1.199 → v2.1.283 ○gone AND v2.1.236 tengu_gzip_request_body_fallback ◇ Requests to the API can be gzip compressed, but both switches default off so nothing compresses.never mined ·never mined never read tengu_gzip_request_body_latch_cleared ◇ The gzip decision now records exactly why it skipped, though compression stays off without a server value.never mined ·never mined never read tengu_halloween ◇ No description from anybody.never mined ·never mined never read tengu_happy_ocean ◇ When you move a session to the cloud from a terminal in auto mode, Claude Code now asks the cloud session to stay in auto modev2.1.296 → v2.1.308 ◆in the build AND v2.1.296 tengu_harbor ◇ Forked skill executions now register as real, trackable background tasksv2.1.138 or earlier → v2.1.308 ◆in the build AND v2.1.290 tengu_harbor_kite ◇ Messages from another session render as an @sender header with indented body instead of raw tags.v2.1.224 → v2.1.308 ◆in the build AND v2.1.296 tengu_harbor_kite_cloud ◇ CLAUDE_CODE_HARBOR_KITE_CLOUD turns on cloud peer access; otherwise a server flag decides and it's off.v2.1.232 → v2.1.308 ◆in the build AND v2.1.248 tengu_harbor_kite_limits ◇ Incoming peer messages now pass rate, duplicate and loop checks before queuing, with drops shown to you.v2.1.242 → v2.1.245 ○gone AND v2.1.223 tengu_harbor_kite_mode_emit ◇ Messages from a session with a mismatched permission mode are held and shown with an approve/deny dialog.v2.1.224 → v2.1.308 ◆in the build AND v2.1.248 tengu_harbor_kite_pacing_off ◇ You can switch off Harbor request pacing with a new environment variable.v2.1.236 → v2.1.308 ◆in the build AND v2.1.248 tengu_harbor_kite_win ◇ Setting CLAUDE_CODE_HARBOR_KITE forces cross-session messaging on, and Windows can now be allowed through.v2.1.229 → v2.1.308 ◆in the build AND v2.1.296 tengu_harbor_ledger ◇ Default []. Built-in allowlist of channel plugins as {marketplace, plugin} entries; a value that fails validation counts as empty. When managed settings give no allowedChannelPlugins, a plugin requested through --channels outside development mode is skipped unless it is on this list. Separately, whatever the managed setting says, an MCP server's claude/channel capability is removed from its reported status unless its plugin is on this list (among other conditions).v2.1.138 or earlier → v2.1.308 ◆in the build AND v2.1.80 tengu_harbor_moth ◇ Default false. Turns on remote recaps: at the end of a turn, in a session that reports metadata changes, Claude Code generates a recap and publishes it as the session's recap metadata, dropping it if a new turn is already running. Further checks must also pass, and CLAUDE_CODE_ENABLE_REMOTE_RECAP, when set, takes precedence over the flag.v2.1.174 → v2.1.308 ◆in the build AND v2.1.174 tengu_harbor_permissions ◇ Default false. Turns on relaying permission prompts to channels: the interactive session registers a channel permission handler, and a tool permission prompt is sent as a notifications/claude/channel/permission_request (tool name, description, input preview) to each connected channel server from the session's --channels list that has registered for it. Tools that require user interaction are not relayed.v2.1.138 or earlier → v2.1.308 ◆in the build AND v2.1.81 tengu_harbor_prism ◇ Default false. When on, the PR status for the current branch is fetched straight from the GitHub REST API (/repos/<owner>/<repo>/pulls?head=...) with a GitHub token, using ETags, redirect handling and backoff on 403/429; when off it runs gh pr view --json number,url,reviewDecision,isDraft,headRefName,state. Either way a null result falls through to a further lookup. The flag also switches the PR-status poller from its fixed 60000 ms interval to an interval computed from terminal focus, time since the last interaction, the failed-fetch streak and the not-modified streak.v2.1.138 or earlier → v2.1.250 ○gone AND v2.1.286 tengu_harbor_willow ◇ In VS Code sessions, unsafe default permission modes from settings are refused and fall back to normal.v2.1.152 → v2.1.308 ◆in the build AND v2.1.296 tengu_harmonic_cook ◇ Messages relayed from a project's coordinator session are handled by a switch that defaults to on, and the model is told when one is waitingv2.1.282 → v2.1.300 ○gone AND v2.1.282 tengu_harp_forepillar ◇ Claude Code now stores state about skills converted into plugins, plus a new flag for keeping something across restartsv2.1.304 → v2.1.308 ◆in the build AND v2.1.304 tengu_harp_pedal ◇ The whiteboard no longer picks a live or solo mode per session and instead follows a single switch controlled remotelyv2.1.304 → v2.1.308 ◆in the build AND v2.1.304 tengu_harp_voatavo ◇ No description from anybody.v2.1.299 → v2.1.308 ◆in the build never read tengu_hawthorn_steeple ◇ Default false. When on, a session keeps a record of tool-result replacements: when the tool results in one message together exceed a per-message size budget, the largest new ones are saved to files and replaced in the conversation, and the same replacements are reapplied on later turns and on resume. Tools without a finite maxResultSizeChars, or that opt out of the aggregate budget, are exempt.v2.1.138 or earlier → v2.1.308 ◆in the build AND v2.1.308 tengu_haze_glass ◇ When only one org memory store is found, it is no longer picked for you automatically.v2.1.210 → v2.1.308 ◆in the build AND v2.1.281 tengu_hazel_lantern ◇ Attachments fetched ahead of time in remote sessions can have their hold renewed, and ones that stall show [attachment did not arrive]v2.1.286 → v2.1.308 ◆in the build AND v2.1.286 tengu_hazel_osprey ◇ Default false. Turns on the context-hint-2026-04-09 beta for main-thread requests that carry first-party betas: once an estimate of savable tokens reaches 20000, the request body gets context_hint with enabled: true and a target_tokens_saved from tengu_hazel_osprey_floor (default 75000, omitted unless above 0). API errors such as 422, 424, 409, 529 or an invalid request turn the hint off again.v2.1.138 or earlier → v2.1.308 ◆in the build AND v2.1.110 tengu_hazel_osprey_floor ◇ Number, code default 75000. While the tengu_hazel_osprey context-hint feature is on for main-thread requests, a request that qualifies for the hint carries context_hint: {enabled: true, target_tokens_saved: <value>}; a value of 0 or less leaves target_tokens_saved out.v2.1.138 or earlier → v2.1.308 ◆in the build AND v2.1.308 tengu_hazel_quire ◇ Claude has the machinery to write dated session log files for auto-memory, but nothing calls it.v2.1.221 → v2.1.259 ○gone AND v2.1.221 tengu_hazel_trellis ◇ Default 3. Maximum subagent nesting depth, used when CLAUDE_CODE_MAX_SUBAGENT_SPAWN_DEPTH is unset; a value that is not an integer of at least 1 uses 3, and the result is cached after the first read. Launching an agent at or past this depth fails with Subagent nesting limit reached (depth N of M), and several agent-tool checks compare against it, some only asking whether it is above 1.v2.1.257 → v2.1.308 ◆in the build AND v2.1.217 tengu_hazy_cosmos ◇ On Linux, Claude Code now refuses to write a file whose path looks like a Mac or Windows user path, and nothing is writtenv2.1.300 → v2.1.308 ◆in the build AND v2.1.300 tengu_hazy_nova ◇ A feature that reports on, off or not loaded now reads a different remote switch, which falls back to offv2.1.288 → v2.1.308 ◆in the build AND v2.1.290 tengu_hazy_peach ◇ Claude Code now checks a new setting that the server can control, and what it does is not yet knownv2.1.295 → v2.1.299 ○gone AND v2.1.295 tengu_headless_fatal_error ◇ Headless runs now print a real reason to stderr and exit 1 instead of dying silently.never mined ·never mined never read tengu_headless_mcp_prewait ◇ Waiting for MCP servers before a headless run is tidied up and logs more detail.never mined ·never mined never read tengu_hearth_member_relay_rows_enabled ◇ Auto mode's safety check gains support for a project member's own session in a shared project, reachable via an environment variable or server flagv2.1.286 → v2.1.308 ◆in the build AND v2.1.286 tengu_hearth_resolved_rows ◇ Extra routing details on relayed messages are stored but ignored unless a server flag turns them on.v2.1.239 → v2.1.308 ◆in the build AND v2.1.242 tengu_heron_brook ◇ On some models Claude will stop spinning up subagents or research workflows unless you actually ask for it.v2.1.150 → v2.1.308 ◆in the build AND v2.1.257 tengu_heron_brook_applied ◇ On some models Claude will stop spinning up subagents or research workflows unless you actually ask for it.never mined ·never mined never read tengu_heron_tallow ◇ Default false. Changes the hint added when the model calls a tool that is not available: for a known tool it reads <tool> is disabled for this session, in subagents as well as here. instead of exists but is not enabled in this context, and for Glob or Grep when Bash is also unavailable it reads <tool> is disabled for this session. CLAUDE_CODE_HERON_TALLOW turns on the same wording.v2.1.217 → v2.1.228 ○gone AND v2.1.217 tengu_herring_clock ◇ Default false. When auto memory is enabled, turns on team memory: a team directory inside the auto-memory directory is loaded into the memory prompt next to the personal one, the team scope can be synced at startup, and writes to it are scanned for secrets (cannot be written to team memory). A non-empty CLAUDE_MEMORY_STORES turns team memory on as well.v2.1.138 or earlier → v2.1.209 ○gone AND v2.1.210 tengu_hidden_frog ◇ A new internal boolean gate was added and defaults to false, with no stated purpose.v2.1.259 → v2.1.308 ◆in the build AND v2.1.269 tengu_hidden_volcano ◇ Claude Code can now offer to move you to a newer model, as a one-line notice or a dialog, when switched on remotelyv2.1.287 → v2.1.308 ◆in the build AND v2.1.287 tengu_hipaa_history_retention_prune ◇ HIPAA-configured organizations get an extra retention pass that prunes stale entries from local history files.v2.1.257 → v2.1.308 ◆in the build AND v2.1.295 tengu_history_picker_scope ◇ An analytics event, not a feature flag. Claude Code logs it when the user cycles the history search picker's scope with the historySearch:cycleScope keybinding (default ctrl+s) through session, project and everywhere. It carries from and to, the old and new scope.never mined ·never mined never read tengu_home_seed_apply ◇ Three events track the path that seeds a fresh home directory, only on accounts where it runs.never mined ·never mined never read tengu_home_seed_hold ◇ Three events track the path that seeds a fresh home directory, only on accounts where it runs.never mined ·never mined never read tengu_home_seed_prompt_resequenced ◇ Your answers to cloud-session prompts are retried with backoff, and you are told if one is lost.never mined ·never mined never read tengu_home_seed_recovery ◇ Three events track the path that seeds a fresh home directory, only on accounts where it runs.never mined ·never mined never read tengu_home_settings_mode_prompt_shown ◇ You now decide separately whether cloud sessions may use your plugins and your local settings.never mined ·never mined never read tengu_home_settings_mode_set ◇ You now decide separately whether cloud sessions may use your plugins and your local settings.never mined ·never mined never read tengu_hook_plugin_metrics ◇ Hook plugin telemetry now resolves plugin id through a more robust lookup.never mined ·never mined never read tengu_hook_prompt_timeout ◇ A prompt hook that times out is now logged and recorded instead of vanishing silently.never mined ·never mined never read tengu_hover_rest ◇ The new storage layer now touches nearly every part of the product but is never actually created.v2.1.221 → v2.1.308 ◆in the build AND v2.1.304 tengu_humble_emerson ◇ No description from anybody.v2.1.287 → v2.1.308 ◆in the build never read tengu_humble_hammock ◇ Two transcript debugging aids, raw tool inputs and exact wire block order, are computed but each sits behind its own gate.v2.1.257 → v2.1.308 ◆in the build AND v2.1.282 tengu_humble_petal ◇ A feature that followed one server-side switch now follows a differently named one, still on by default when the server sends nothingv2.1.285 → v2.1.308 ◆in the build AND v2.1.285 tengu_humble_wand ◇ Auto mode now re-checks artifact facts to catch stale 'allow' verdicts from the server classifierv2.1.271 → v2.1.308 ◆in the build AND v2.1.271 tengu_humble_zephyr ◇ Task-management reminders are now skipped in conversations that used the Brief toolv2.1.277 → v2.1.308 ◆in the build AND v2.1.277 tengu_humming_locket ◇ Someone connected through Remote Control can now run /effort, /model, /rename and /usage straight away while Claude is mid-replyv2.1.290 → v2.1.308 ◆in the build AND v2.1.290 tengu_humming_star ◇ A Remote Control inbox message that says it is from the owner is treated as from an unknown sender unless the server confirmed itv2.1.299 → v2.1.308 ◆in the build AND v2.1.299 tengu_humming_sundae ◇ Claude Code can now remind Claude about remote notifications that have waited longer than a set number of minutesv2.1.295 → v2.1.308 ◆in the build AND v2.1.295 tengu_hushed_kestrel ◇ When a routed agent's single clean handback goes to that same agent, Claude Code can use a short pointer notice insteadv2.1.285 → v2.1.308 ◆in the build AND v2.1.285 tengu_hushed_lantern ◇ Claude in Chrome tool calls are now denied straight away with 'no browser connected' when no browser extension is attachedv2.1.299 → v2.1.308 ◆in the build AND v2.1.299 tengu_hushed_lark ◇ Silent-turn reminder and artifact room environment variables are now recognised by name in the settings allowlist.v2.1.236 → v2.1.308 ◆in the build AND v2.1.290 tengu_hushed_lark_text ◇ Silent-turn reminder and artifact room environment variables are now recognised by name in the settings allowlist.v2.1.238 → v2.1.308 ◆in the build AND v2.1.290 tengu_hushed_plover ◇ If a permission prompt times out unanswered, later subagent requests in the same turn are held back instead of asking againv2.1.304 → v2.1.308 ◆in the build AND v2.1.304 tengu_ide_rc_auto_enable ◇ Startup telemetry now reports Remote Control availability and an old experiment field was dropped.v2.1.203 → v2.1.308 ◆in the build AND v2.1.285 tengu_idempotent_wolf ◇ New instructions tell some models that reasoning effort does not reduce how much of a request to finish, and not to check in before big tasksv2.1.293 → v2.1.308 ◆in the build AND v2.1.293 tengu_idle_amber_finch ◇ Default false. When on, Claude Code stops offering an upgrade to Max: the usage-limit menu drops its Upgrade to Max option, the shared upgrade-availability check returns false (so, for example, the five-hour limit message no longer suggests /upgrade), and the startup subscription check no longer reports an existing Claude Max subscription.v2.1.138 or earlier → v2.1.308 ◆in the build AND v2.1.308 tengu_image_output ◇ No description from anybody.never mined ·never mined never read tengu_image_resize ◇ Image resizing now uses a separate, adjustable threshold for rejecting images instead of reusing the resize target sizenever mined ·never mined never read tengu_image_resize_failed ◇ Image resizing now uses a separate, adjustable threshold for rejecting images instead of reusing the resize target sizenever mined ·never mined never read tengu_image_resize_fallback ◇ Image resizing now uses a separate, adjustable threshold for rejecting images instead of reusing the resize target sizenever mined ·never mined never read tengu_image_webp_direct ◇ No description from anybody.never mined ·never mined never read tengu_immediate_command_executed ◇ Custom slash-command names are now hidden from telemetry, replaced with 'custom'.never mined ·never mined never read tengu_immediate_model_command ◇ /help, /theme and /add-dir now respond instantly; /config and /advisor do only if a remote flag turns on.v2.1.138 or earlier → v2.1.241 ○gone AND v2.1.242 tengu_immutable_tide ◇ A check that runs only when a server value is set now also returns false for a result whose reason is that it did not runv2.1.271 → v2.1.308 ◆in the build AND v2.1.292 tengu_import ◇ Importing another agent's config now parses TOML natively, reachable only through the gated import path.v2.1.213 → v2.1.308 ◆in the build AND v2.1.265 tengu_import_apply ◇ /import no longer repeats the same warning multiple times in its 'Review these:' listv2.1.246 → v2.1.247 ○gone never read tengu_indexed_corbato ◇ Conversations can now carry coordinator/worker context and named context sections, behind an unreleased gatev2.1.268 → v2.1.308 ◆in the build AND v2.1.268 tengu_indexed_pearl ◇ No description from anybody.v2.1.300 → v2.1.308 ◆in the build never read tengu_init ◇ Startup telemetry now notes whether your project uses git, Docker, Nix, Bazel, CI configs and similar.never mined ·never mined never read tengu_inline_tools_announced ◇ An analytics event, not a feature flag. Claude Code logs it while building the deferred_tools_delta for a turn when it announces tool definitions inline and at least one count is above zero; the fields are definitions, replacements, pointers and withdrawals.never mined ·never mined never read tengu_input_prompt ◇ The effort_level recorded with tengu_input_prompt now also accounts for active permission layersnever mined ·never mined never read tengu_input_slash_invalid ◇ A slash command can now be refused by a policy attached to it, with the reason shown.never mined ·never mined never read tengu_iridescent_boot ◇ rm -r on a path made entirely of $(...) or backtick output now always asks, and permission rules cannot auto-allow itv2.1.281 → v2.1.308 ◆in the build AND v2.1.305 tengu_iridescent_crystal ◇ Auto-mode's safety classifier gained a new 'arbiter' mode that can skip some server checksv2.1.269 → v2.1.308 ◆in the build AND v2.1.281 tengu_iridescent_gem ◇ New sampled telemetry records detailed hashes around cache-prefix breaks.v2.1.261 → v2.1.308 ◆in the build AND v2.1.308