Feature flags, all 2,081.

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_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_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_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_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_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_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_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_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_jazzy_bird ◇ A gate that would skip commit and PR attribution text and suppress bridge conflict recovery defaults to off.v2.1.257 → v2.1.258 ○gone AND v2.1.257 tengu_joyful_globe ◇ No description from anybody.v2.1.259 → v2.1.308 ◆in the build AND v2.1.308 tengu_kairos_input_needed_push ◇ Default false. Shows the Push when actions required toggle (the inputNeededNotifEnabled setting) in the settings list and the matching row in the Notifications panel; both places also need a further condition to hold.v2.1.138 or earlier → v2.1.308 ◆in the build AND v2.1.110 tengu_kairos_loop_dynamic ◇ Checkpoints now record self-scheduling wake counts and open artifact watches, though the loop itself is off.v2.1.138 or earlier → v2.1.247 ○gone AND v2.1.248 tengu_kairos_loop_prompt ◇ The Kairos loop prompt is hardcoded off and its remote switch is gone from the build.v2.1.138 or earlier → v2.1.247 ○gone AND v2.1.248 tengu_kairos_push_notifications ◇ A tool letting Claude push notifications to you stays off by default.v2.1.138 or earlier → v2.1.308 ◆in the build AND v2.1.248 tengu_keen_lantern ◇ One check now reads its remote switch under a fixed name and still falls back to on when the server sends nothingv2.1.288 → v2.1.308 ◆in the build AND v2.1.288 tengu_keybinding_customization_release ◇ Your custom keybindings can load in the background at startup so they're ready sooner.v2.1.138 or earlier → v2.1.308 ◆in the build AND v2.1.225 tengu_lapis_anchor_user_turn ◇ Claude now gets a running count of tokens left by default, re-anchored on each of your prompts.v2.1.202 → v2.1.308 ◆in the build AND v2.1.236 tengu_lazy_dove ◇ Claude Code now counts turns since a person last wrote and can act once a limit is reached, but this is off unless switched on remotelyv2.1.296 → v2.1.308 ◆in the build AND v2.1.296 tengu_lexical_crystal ◇ No description from anybody.v2.1.287 → v2.1.308 ◆in the build AND v2.1.308 tengu_lexical_glade ◇ A warning for switching account or organization mid-conversation is built but held behind the tengu_lexical_glade flagv2.1.281 → v2.1.308 ◆in the build AND v2.1.307 tengu_lively_waffle ◇ Subagents running in 'auto' mode may need to report back via a new SubagentHandback tool instead of plain text, gated by an unread flagv2.1.267 → v2.1.308 ◆in the build AND v2.1.271 tengu_log_datadog_events ◇ Default false. When on, each analytics event that passes sampling is also sent to Datadog, in addition to the first-party event logger; Datadog only accepts it with the first-party API provider and for event names on a built-in list. A datadog: true entry in the remote sink-disable config turns it off regardless.v2.1.138 or earlier → v2.1.308 ◆in the build AND v2.0.36 tengu_loop_noop_fold ◇ Several artifact feature switches were renamed and re-gated, discarding any prior rollout.v2.1.198 → v2.1.241 ○gone AND v2.1.242 tengu_luminous_seal ◇ Two new server-controlled switches govern whether reset and rate-limit events get forwarded.v2.1.224 → v2.1.308 ◆in the build AND v2.1.224 tengu_luminous_whistle ◇ Reasoning text from a non-Claude model can survive a model switch, off unless enabled.v2.1.251 → v2.1.278 ○gone AND v2.1.308 tengu_madrone_spindle ◇ Artifact comment replies get grouped wakes and a stand-down when everything was already answered, both switched off.v2.1.234 → v2.1.308 ◆in the build AND v2.1.234 tengu_marble_whisper ◇ No description from anybody.never mined ·never mined AND v2.1.70 tengu_marbled_teal ◇ Effort levels can be labelled with relative cost like ~2x, hidden behind an off gate.v2.1.224 → v2.1.308 ◆in the build AND v2.1.224 tengu_mcp_claudeai_eligibility_gate ◇ Default false. When on, a claude.ai connector (claudeai-proxy MCP server) whose config has eligible: false is not connected and is reported as needs-auth.v2.1.212 → v2.1.308 ◆in the build AND v2.1.308 tengu_mcp_legacy_url_elicitation ◇ MCP server configs accept a new bareElicitationCapability option that gives the server the plain elicitation capability instead of the legacy URL onev2.1.281 → v2.1.308 ◆in the build AND v2.1.287 tengu_mcp_protocol_negotiation_ccr ◇ The ccr-proxy MCP transport can negotiate protocol versions, but stays on legacy unless a flag is turned on.v2.1.236 → v2.1.308 ◆in the build AND v2.1.299 tengu_mcp_protocol_negotiation_claudeai ◇ A newer MCP protocol handshake exists per transport but every gate falls back to legacy here.v2.1.221 → v2.1.308 ◆in the build AND v2.1.295 tengu_mcp_protocol_negotiation_http ◇ A newer MCP protocol handshake exists per transport but every gate falls back to legacy here.v2.1.221 → v2.1.308 ◆in the build AND v2.1.295 tengu_mcp_protocol_negotiation_stdio ◇ A newer MCP protocol handshake exists per transport but every gate falls back to legacy here.v2.1.221 → v2.1.308 ◆in the build AND v2.1.292 tengu_mcp_singleton_unwrap ◇ MCP tool calls and resumed tasks now hold a guard on connection error tracking for as long as each call runsv2.1.141 → v2.1.308 ◆in the build AND v2.1.287 tengu_mcp_stateless_skip_init ◇ Default true. When on, a claude.ai connector (claudeai-proxy) with stateless: true connects statelessly: MCP initialize is answered from its cached init response when that parses and carries a supported protocol version (otherwise a real initialize runs), and when tengu_mcp_discover_projection_prior is not off, server/discover can be resolved from a cached projection. It also turns on the anthropic-mcp-client-capabilities and MCP-Protocol-Version request headers even when tengu_mcp_discover_projection_prior is off.v2.1.169 → v2.1.308 ◆in the build AND v2.1.308 tengu_mcp_step_up_auth_dialog ◇ Claude Code can offer to sign in to an MCP server again with wider permissions when a tool call is refused for missing scopev2.1.281 → v2.1.308 ◆in the build AND v2.1.288 tengu_melodic_ember ◇ A check in the plugin and hook area now depends on a remote switch, with a new step for VS Code making sure hook registration is not pendingv2.1.299 → v2.1.308 ◆in the build AND v2.1.304 tengu_misty_twilight ◇ New helper can abort leftover elicitation watchdogs, controlled by the tengu_misty_twilight flagv2.1.280 → v2.1.308 ◆in the build AND v2.1.280 tengu_mocha_barista ◇ /schedule now always documents one-time runs and how to re-arm one that already fired.v2.1.138 or earlier → v2.1.245 ○gone AND v2.1.246 tengu_modular_mochi ◇ Destructive MCP calls under remote auto mode can now be allowed by either the existing gate or a new flag.v2.1.259 → v2.1.308 ◆in the build AND v2.1.295 tengu_moonlit_panda ◇ Cloud sessions could delay starting language servers until needed, but the gate is off here.v2.1.246 → v2.1.308 ◆in the build AND v2.1.246 tengu_mossy_lantern ◇ Tools added or removed mid-conversation can be surfaced to the model, but the beta capability is gated off by default.v2.1.257 → v2.1.268 ○gone AND v2.1.308 tengu_native_cursor ◇ Default false. When true, Claude Code uses the terminal's native cursor, the same as setting CLAUDE_CODE_NATIVE_CURSOR, provided the DECSTBM renderer is not in use; the result is cached for the process. CLAUDE_CODE_ACCESSIBILITY and one other internal condition turn the native cursor on without this flag. On Windows the same switches also feed a second terminal setup check that does not look at the DECSTBM renderer.v2.1.163 → v2.1.308 ◆in the build AND v2.1.308 tengu_onyx_sluice ◇ Artifact comments could be fetched server-side instead of through the frame host, but that switch is off.v2.1.228 → v2.1.308 ◆in the build AND v2.1.281 tengu_orford_ness ◇ Default false. Turns on sending error reports (logged errors, unhandled rejections and uncaught exceptions) to Datadog error tracking, with a per-process report cap. It is checked last: reporting also needs DISABLE_ERROR_REPORTING unset, the first-party API provider, a Claude Code version of at least 2.1.193, the allow_error_reporting policy check passing, and other internal checks.v2.1.191 → v2.1.274 ○gone AND v2.1.308 tengu_penguins_enabled ◇ No description from anybody.never mined ·never mined AND v2.1.36 tengu_peppy_riddle ◇ A check that matched MCP servers against a remote list can now read a single on/off switch instead, depending on how it is calledv2.1.300 → v2.1.304 ○gone AND v2.1.300 tengu_peppy_zephyr ◇ Tool token counts for context analysis can come from the server instead of local estimates, but the gate is off.v2.1.257 → v2.1.276 ○gone AND v2.1.261 tengu_per_turn_effort ◇ Per-family capability gates now use a name-to-flag table instead of matching skill and tool name strings.never mined ·never mined AND v2.1.308 tengu_pewter_brook ◇ Accepting fullscreen now runs as a trial, and two crashed launches disable it on that machine.v2.1.138 or earlier → v2.1.308 ◆in the build AND v2.1.239 tengu_pewter_orchard ◇ No description from anybody.v2.1.210 → v2.1.231 ○gone AND v2.1.308 tengu_pewter_summit ◇ No description from anybody.v2.1.145 → v2.1.308 ◆in the build AND v2.1.308 tengu_plain_sorrel ◇ When checking a shell command before running it, Claude Code now handles assignment-style arguments to dd under a separate rulev2.1.307 → v2.1.308 ◆in the build AND v2.1.307 tengu_plucky_orchard ◇ Tool definitions could survive context loss and credential changes, but both halves are off.never mined ·never mined AND v2.1.267 tengu_plugin_official_mkt_git_fallback ◇ Plugin installs can now fall back to GitHub over HTTPS, and bulk marketplace refreshes report skipped built-in marketplaces separatelyv2.1.138 or earlier → v2.1.308 ◆in the build AND v2.1.290 tengu_plum_vx3 ◇ Web search tool's model choice now goes through a feature-flag check instead of always using the main modelv2.1.138 or earlier → v2.1.308 ◆in the build AND v2.1.269 tengu_polished_lagoon ◇ One session could inject a coordination notice into another, but three separate gates keep it dark.v2.1.234 → v2.1.308 ◆in the build AND v2.1.308 tengu_polished_salamander ◇ A new server switch lets a resumed session keep open a message that was waiting on a permission answer when the turn was interruptednever mined ·never mined AND v2.1.288 tengu_polished_tulip ◇ The git and commit instructions Claude receives can now include an extra line about checks before shipping, depending on which skills are loadedv2.1.286 → v2.1.308 ◆in the build AND v2.1.296 tengu_print_engine_loop ◇ The env var controlling print and engine loop behavior was renamed to CLAUDE_CODE_HUMBLE_HAMMOCK and still defaults to off.v2.1.247 → v2.1.252 ○gone AND v2.1.257 tengu_projects_compact_thread_rule_enabled ◇ Auto mode's permission check can use a 'compact' view of the conversation, off unless switched on remotely or by an environment variablev2.1.299 → v2.1.308 ◆in the build AND v2.1.299 tengu_prompt_cache_diagnostics ◇ Cache diagnostics know when the baseline came from disk and skip misleading difference reports.v2.1.138 or earlier → v2.1.245 ○gone AND v2.1.246 tengu_quartz_pipit ◇ Default true. When a main-thread turn under an active /goal ends in a way Claude Code classifies as unrecoverable (a context limit such as prompt_too_long or blocking_limit), the goal is cleared, its Stop hook removed, and the warning Goal cleared after an unrecoverable error (...) is shown. When false, that case returns without clearing the goal.v2.1.234 → v2.1.308 ◆in the build AND v2.1.308 tengu_quartz_thimble ◇ Plugin eval now writes report.html and publishes it to claude.ai unless you pass --no-publish.v2.1.224 → v2.1.308 ◆in the build AND v2.1.269 tengu_quiet_larkspur ◇ Some allowed shell commands followed by a NAME=value argument no longer fail the permission check as an unsafe variable settingv2.1.285 → v2.1.308 ◆in the build AND v2.1.307 tengu_quiet_locket ◇ Behind a server-side switch, cloud sessions take your deny and ask rules from your own settings and keep allow rules only if both sides have themnever mined ·never mined AND v2.1.288 tengu_report_findings_tool ◇ Two server-side switches that steered review behaviour are gone, so reviews no longer vary by remote config.v2.1.196 → v2.1.231 ○gone AND v2.1.232 tengu_retire_chat_relay_artifact_backstop ◇ A chat relay flag no longer does anything; entrypoint alone decides that path.v2.1.203 → v2.1.247 ○gone AND v2.1.248 tengu_rewind_first_message ◇ Default false. Changes the SDK rewind_conversation control request: when false, the rewind is anchored at the nearest assistant message before the target and fails with no preceding assistant if there is none. When true, it is anchored at the nearest user or assistant message before the target, that error is not returned (so the first message can be the target), and the anchor is persisted with { rewound: true }.v2.1.191 → v2.1.245 ○gone AND v2.1.308 tengu_sable_thrush ◇ Claude Code now spots server-tagged narration thinking, but the display change stays off by default.v2.1.240 → v2.1.250 ○gone AND v2.1.251 tengu_saddle_lantern ◇ The memory prompt variant and the skill-suggestion rollout are now fields on the process config.v2.1.260 → v2.1.306 ○gone AND v2.1.248 tengu_saffron_anchor ◇ A prompt-and-context experiment switch was renamed and four other switches disappeared entirely.v2.1.179 → v2.1.245 ○gone AND v2.1.246 tengu_saffron_wren ◇ Goals now nudge Claude every 30 minutes when background shells or agents are blocking evaluation.v2.1.234 → v2.1.308 ◆in the build AND v2.1.269 tengu_sdk_mcp_manifests ◇ SDK-registered in-process MCP servers can now skip a round-trip handshake at session startv2.1.265 → v2.1.308 ◆in the build AND v2.1.288 tengu_sessions_elevated_auth_enforcement ◇ Cloud sessions get told outright they cannot message peers that require a trusted device.v2.1.138 or earlier → v2.1.308 ◆in the build AND v2.1.282 tengu_sharded_beacon ◇ Subscribers whose sign-in token has expired get a dedicated error when Claude Code times out waiting to refresh itv2.1.248 → v2.1.308 ◆in the build AND v2.1.290 tengu_shoji_engine ◇ No description from anybody.v2.1.193 → v2.1.202 ○gone AND v2.1.308 tengu_silly_waterfall ◇ A gated fix can now remove stale leftover transcript entries when resuming a headless sessionv2.1.271 → v2.1.308 ◆in the build AND v2.1.295 tengu_slate_anchor ◇ Some feature flags now stay fixed for the whole session, including a new one that is on.never mined ·never mined AND v2.1.248 tengu_slate_fern ◇ Default false. Turns on the unused plugin check: user-installed marketplace plugins not used for at least days days and sessions sessions feed the plugin-disuse-review startup tip (You haven't used the ... plugin in a while) and the days-since-last-use shown in the plugin list. Plugins that provide LSP servers, themes, output styles, monitors or workflows are never flagged, and the check is off when the strictKnownMarketplaces policy is set. true uses 14 days and 10 sessions; an object can set numeric days and sessions, each falling back to those values.v2.1.172 → v2.1.186 ○gone AND v2.1.308 tengu_slate_ibis ◇ Explore and plan agents are now always on unless you set the disable environment variable yourself.v2.1.217 → v2.1.220 ○gone AND v2.1.221 tengu_slate_kestrel ◇ Default false. When on, a claude.ai OAuth login with no known subscription type (null) counts as eligible for organization policy limits; otherwise an OAuth login is eligible only on an enterprise or team subscription. A login with an API key is eligible either way, and the other checks (first-party API, the token's scope) still apply.v2.1.138 or earlier → v2.1.148 ○gone AND v2.1.308 tengu_slate_lantern ◇ A new --watch-artifact flag follows a claude.ai artifact for new versions and comments, but needs account access.v2.1.198 → v2.1.308 ◆in the build AND v2.1.274 tengu_slate_lantern_ember ◇ Re-armed artifact watches could reuse an existing connection instead of starting over, but the flag is off.v2.1.236 → v2.1.308 ◆in the build AND v2.1.257 tengu_slate_meadow ◇ Default false. When true, enables the background-agents fleet gate (isAgentsFleetEnabled) and adds the built-in Claude agent to the built-in agent list, unless agent view is disabled: by CLAUDE_CODE_DISABLE_AGENT_VIEW, the disableAgentView setting, or a third-party API provider. Both are also on without the flag when Claude Code runs in essential-traffic-only mode (for example with CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC set).v2.1.138 or earlier → v2.1.140 ○gone AND v2.1.308 tengu_slate_quill ◇ Default false. One of the gates for the DesignSync tool and the /design, /design-consent, /design-revoke and /design-login commands. They are off when the allow_design_sync policy check fails or another internal check disables them, on when a further internal check passes, and otherwise follow this flag.v2.1.160 → v2.1.245 ○gone AND v2.1.308 tengu_slate_ribbon ◇ No description from anybody.never mined ·never mined AND v2.1.110 tengu_slim_subagent_claudemd ◇ Default true. When on, subagents whose definition sets omitClaudeMd (the built-in Explore and Plan agents in this version) start without the CLAUDE.md content in their user context, unless the caller passes its own user context. When off, they receive the CLAUDE.md content like other subagents.v2.1.138 or earlier → v2.1.195 ○gone AND v2.1.308 tengu_smooth_forest ◇ Rate-limit menu gains a new promotional "limit wall" option pointing to a claim pagev2.1.275 → v2.1.304 ○gone AND v2.1.275 tengu_sorrel_trellis ◇ Claude can watch artifact comment threads and post replies or edits, off unless you set the env var.v2.1.221 → v2.1.308 ◆in the build AND v2.1.246 tengu_sorrel_trellis_lintel ◇ A comment sent moments before a watch starts could get a reply, but the flag is off.v2.1.236 → v2.1.267 ○gone AND v2.1.308 tengu_sotto_voce ◇ No description from anybody.never mined ·never mined AND v2.1.66 tengu_steady_panda ◇ When a session is set up, Claude Code's first reply now waits, up to a limit, for the session's start-up work to finishv2.1.287 → v2.1.308 ◆in the build AND v2.1.287 tengu_stellar_drift ◇ 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 v2.1.281 tengu_still_kestrel ◇ Stored credentials could survive a token change, but not in this build.never mined ·never mined AND v2.1.248 tengu_streaming_tool_execution2 ◇ Default false. Turns on streaming tool execution in the query loop: tool calls are handed to an executor as their tool_use blocks arrive while the model's response is still streaming, completed results are emitted during the stream, and the remaining results are collected after it ends. The executor is discarded and recreated when a response is retried or restarted.v2.1.138 or earlier → v2.1.153 ○gone AND v2.1.308 tengu_sub_nomdrep_q7k ◇ No description from anybody.never mined ·never mined AND v2.1.91 tengu_surreal_dali ◇ Claude can read a routine's past runs and detailed run log, but a server flag gates it.v2.1.138 or earlier → v2.1.250 ○gone AND v2.1.251