Environment variables, all 1,448.

CLAUDE_BRIDGE_BASE_URL ◇ Admin and managed settings now merge their env blocks instead of one replacing the other; you can opt out.v2.1.160 → v2.1.308 ◆in the build CLAUDE_BRIDGE_OAUTH_TOKEN ◇ Admin and managed settings now merge their env blocks instead of one replacing the other; you can opt out.v2.1.160 → v2.1.308 ◆in the build CLAUDE_BRIDGE_REATTACH_GROUPING ◇ Part of the Remote Control reattach handoff Claude Code puts in a child process's environment when it relaunches or respawns a session that had a bridge session. Holds that bridge session's grouping ID. The new process uses it as the reattached session's grouping ID, only when CLAUDE_BRIDGE_REATTACH_SESSION is also set and the reattach is not vetoed, and removes it from its environment after reading.v2.1.206 → v2.1.308 ◆in the build CLAUDE_BRIDGE_REATTACH_NO_BACKFILL ◇ Part of the Remote Control reattach handoff Claude Code puts in a child process's environment when it relaunches or respawns a session that had a bridge session. Claude Code sets it to 1 when the conversation's recorded bridge state says not to backfill history. When it is truthy (1, true, yes, on) and CLAUDE_BRIDGE_REATTACH_SESSION is set, the new process reattaches with history channels suppressed. It is removed from the environment after reading.v2.1.227 → v2.1.308 ◆in the build CLAUDE_BRIDGE_REATTACH_OUTBOUND_ONLY ◇ One resolver now decides whether Remote Control starts and in which mode, weighing flags, settings and env vars.v2.1.147 → v2.1.308 ◆in the build CLAUDE_BRIDGE_REATTACH_OWNER_ACCT ◇ Resuming a session under a different account starts fresh instead of replaying the old history.v2.1.227 → v2.1.308 ◆in the build CLAUDE_BRIDGE_REATTACH_OWNER_ORG ◇ Part of the Remote Control reattach handoff Claude Code puts in a child process's environment when it relaunches or respawns a session that had a bridge session. Holds the organization UUID of the conversation's recorded owner, alongside the account UUID in CLAUDE_BRIDGE_REATTACH_OWNER_ACCT. When both the handoff and the current login have an account, the new process compares account and organization; on a mismatch it vetoes the reattach and mints a fresh bridge session with history channels suppressed. It is removed from the environment after reading.v2.1.227 → v2.1.308 ◆in the build CLAUDE_BRIDGE_REATTACH_SEQ ◇ Part of the Remote Control reattach handoff Claude Code puts in a child process's environment when it relaunches or respawns a session that had a bridge session. Holds the last message sequence number of the bridge session named by CLAUDE_BRIDGE_REATTACH_SESSION, as a decimal integer, and is only written when greater than 0. The new process parses it as the sequence number to reattach at, only when CLAUDE_BRIDGE_REATTACH_SESSION is set, and removes it from the environment after reading.v2.1.138 or earlier → v2.1.308 ◆in the build CLAUDE_BRIDGE_REATTACH_SESSION ◇ One resolver now decides whether Remote Control starts and in which mode, weighing flags, settings and env vars.v2.1.138 or earlier → v2.1.308 ◆in the build CLAUDE_BRIDGE_SESSION_INGRESS_URL ◇ Admin and managed settings now merge their env blocks instead of one replacing the other; you can opt out.v2.1.160 → v2.1.308 ◆in the build CLAUDE_BRIDGE_USE_CCR_V2 ◇ In bridge (Remote Control) mode, when 1, true, yes or on, or when the work item has use_code_sessions: true, each new session is registered as a CCR v2 worker at <apiBaseUrl>/v1/code/sessions/<id>/worker/register (two attempts, 2 seconds apart) instead of using the session_ingress WebSocket URL. If both attempts fail, the work item is stopped and no session is spawned; on success the child process gets CLAUDE_CODE_USE_CCR_V2=1 and CLAUDE_CODE_WORKER_EPOCH.v2.1.138 or earlier → v2.1.202 ○gone CLAUDE_BYTE_STREAM_IDLE_TIMEOUT_MS Timeout in milliseconds for the byte-level streaming idle watchdog; when set, it takes precedence over CLAUDE_STREAM_IDLE_TIMEOUT_MS for that watchdog and leaves the event-level watchdog unchanged. Claude Code clamps this variable to between 10 seconds and 30 minutes. Requires Claude Code v2.1.210 or laterv2.1.157 → v2.1.308 ◆in the build CLAUDE_CHROME_CLASSIFIER_FLOOR ◇ The terminal tab status toggle, Claude API tip and Chrome classifier default only moved; their gate fallbacks are unchangedv2.1.187 → v2.1.308 ◆in the build CLAUDE_CHROME_PAIRED_DEVICE_ID ◇ A new CLAUDE_CHROME_PAIRED_DEVICE_ID env var pins which paired browser device Claude in Chrome connects tov2.1.268 → v2.1.308 ◆in the build CLAUDE_CHROME_PERMISSION_MODE ◇ Claude Code now knows safe values for three Chrome variables, can clear them, and keeps a new unattended-tools variable out of child processesv2.1.138 or earlier → v2.1.308 ◆in the build CLAUDE_CHROME_TAB_GROUP_KEY ◇ Claude Code now reads CLAUDE_CODE_CONFIG_PROBE, CLAUDE_CODE_ARTIFACT_SHARE and CLAUDE_CHROME_TAB_GROUP_KEY when it startsv2.1.286 → v2.1.308 ◆in the build CLAUDE_CLIENT_PRESENCE_FILE Path to a file that an external tool, such as a screen-lock listener, creates when you unlock your screen and deletes when you lock it. While the file exists, Claude Code skips Remote Control mobile push notifications, so you stop getting pushes while you are actively using the computer. When the file is absent or unreadable, notifications are sent as normal. Claude Code checks the file once per push-triggering event rather than polling it. Requires Claude Code v2.1.181 or laterv2.1.181 → v2.1.308 ◆in the build CLAUDE_CODE_3P_PROBE_WROTE_HAIKU_DEFAULT ◇ If your cloud account cannot use a model, Claude Code can step down to the next one; CLAUDE_CODE_DISABLE_MODEL_ACCESS_FALLBACK turns this offv2.1.285 → v2.1.308 ◆in the build CLAUDE_CODE_3P_PROBE_WROTE_OPUS_DEFAULT ◇ Internal marker on third-party providers. When Claude Code itself writes ANTHROPIC_DEFAULT_OPUS_MODEL into its environment (the model availability probe on Bedrock, Vertex or Mantle, or a move down the model fallback ladder), it sets this variable to the same model ID. An ANTHROPIC_DEFAULT_OPUS_MODEL equal to it is treated as Claude Code's own choice rather than a user pin: it does not count as pinned, the probe does not skip it, and the fallback ladder may move it.v2.1.207 → v2.1.308 ◆in the build CLAUDE_CODE_3P_PROBE_WROTE_SONNET_DEFAULT ◇ On third-party providers, Haiku model names are no longer skipped by the check that sets up the Sonnet default modelv2.1.207 → v2.1.308 ◆in the build CLAUDE_CODE_3P_SEEDED_OPUS_DEFAULT ◇ If your cloud account cannot use a model, Claude Code can step down to the next one; CLAUDE_CODE_DISABLE_MODEL_ACCESS_FALLBACK turns this offv2.1.285 → v2.1.308 ◆in the build CLAUDE_CODE_3P_SEEDED_SONNET_DEFAULT ◇ If your cloud account cannot use a model, Claude Code can step down to the next one; CLAUDE_CODE_DISABLE_MODEL_ACCESS_FALLBACK turns this offv2.1.285 → v2.1.308 ◆in the build CLAUDE_CODE_ACCESSIBILITY Set to 1 to keep the native terminal cursor visible and disable the inverted-text cursor indicator. Allows screen magnifiers like macOS Zoom to track cursor positionv2.1.138 or earlier → v2.1.308 ◆in the build CLAUDE_CODE_ACCOUNT_TAGGED_ID ◇ Overrides the user.account_id attribute Claude Code attaches to its OpenTelemetry data. It applies only when an account UUID is known and OTEL_METRICS_INCLUDE_ACCOUNT_UUID is not turned off (default on); when unset or empty, user.account_id is derived from the account UUID as a tagged ID of the form user_01....v2.1.138 or earlier → v2.1.308 ◆in the build CLAUDE_CODE_ACCOUNT_UUID ◇ The device bridge is skipped entirely when the host's account UUID doesn't match the stored one.v2.1.138 or earlier → v2.1.308 ◆in the build CLAUDE_CODE_ACT_DONT_REDERIVE ◇ Three existing feature gates moved position; their defaults are unchanged.v2.1.162 → v2.1.308 ◆in the build CLAUDE_CODE_ACTION ◇ Marks a run as coming from the Claude Code GitHub Action. A truthy value (1, true, yes or on) makes Claude Code set CLAUDE_CODE_ENTRYPOINT to claude-code-github-action, but only when CLAUDE_CODE_ENTRYPOINT is not already set and the command is not claude mcp serve. It is also reported as isClaudeCodeAction in the environment details attached to telemetry.v2.1.138 or earlier → v2.1.308 ◆in the build CLAUDE_CODE_ADDITIONAL_DIRECTORIES_CLAUDE_MD Set to 1 to load memory files from directories specified with --add-dir. Loads CLAUDE.md, .claude/CLAUDE.md, .claude/rules/*.md, and CLAUDE.local.md. By default, additional directories do not load memory filesv2.1.138 or earlier → v2.1.308 ◆in the build CLAUDE_CODE_ADDITIONAL_PROTECTION ◇ When truthy (1, true, yes or on), Claude Code adds the header x-anthropic-additional-protection: true to the default headers of the Anthropic API clients it creates.v2.1.138 or earlier → v2.1.308 ◆in the build CLAUDE_CODE_ADOPT_UNDERIVABLE_PARKED_PERMISSION ◇ A resumed headless session could hand you back a permission prompt left unanswered, but three variables must be set.v2.1.238 → v2.1.306 ○gone CLAUDE_CODE_AGENT ◇ Set by Claude Code to the value of --agent <agent> when that option is given. It is read when the session registers itself in the concurrent-sessions registry, where it is stored as the agent field of the session's pid-file record.v2.1.138 or earlier → v2.1.308 ◆in the build CLAUDE_CODE_AGENT_COST_STEER ◇ Forces the Agent tool's cost steer: a true value (1, true, yes, on) turns it on, a false value (0, false, no, off) turns it off, and otherwise it is on when the subscription type is pro and tengu_willow_prism is true. When on, the Agent tool prompt gains the paragraph Do not spawn agents unless the user asks. and drops the bullet about launching several agents in one message, and the initial agent listing omits its concurrency note.v2.1.138 or earlier → v2.1.139 ○gone CLAUDE_CODE_AGENT_LIST_IN_MESSAGES ◇ Where the Agent tool's list of agent types goes. A true value (1, true, yes, on) moves the list out of the tool description into agent_listing_delta attachments in the conversation, and the description says the types are listed in <system-reminder> messages; a false value (0, false, no, off) keeps the list in the description. Unset or any other value defers to the tengu_agent_list_attach flag, and a truthy CLAUDE_CODE_DISABLE_ATTACHMENTS or CLAUDE_CODE_SIMPLE keeps the list in the description whatever this is set to.v2.1.138 or earlier → v2.1.177 ○gone CLAUDE_CODE_AGENT_PROGRESS_SUMMARIES ◇ Claude Code now reads three new environment variables: agent progress summaries, a sleep-compaction delay and a resume hook waitv2.1.296 → v2.1.308 ◆in the build CLAUDE_CODE_AGENT_PROXY_GH_SHIM ◇ The agent/tool proxy can now add extra git config includes and prepend a directory to PATHv2.1.202 → v2.1.308 ◆in the build CLAUDE_CODE_AGENT_PROXY_GIT_CONFIG ◇ Turns on governed git routing in Claude Code's agent proxy; Claude Code sets it to 1 in a remote session's environment when that session's tool config enables gitConfig. When truthy and the agent proxy starts (a remote session with CCR_AGENT_PROXY_ENABLED), it appends a marked governed-git block to the git config file named by GIT_CONFIG_GLOBAL (skipped with a warning if that is unset), routing HTTPS git traffic for the governed hosts (CLAUDE_CODE_AGENT_PROXY_GIT_HOSTS, default github.com) through the local proxy with its CA bundle and git credential prompts disabled. Like CLAUDE_CODE_AGENT_PROXY_GH_SHIM, it also puts the proxy in its tool-scoped mode, and a settings file's env block cannot set it.v2.1.202 → v2.1.308 ◆in the build CLAUDE_CODE_AGENT_PROXY_GIT_HOSTS ◇ A new CLAUDE_CODE_AGENT_PROXY_GIT_HOSTS variable lists the git hosts a session sends through the agent proxy, instead of just onev2.1.283 → v2.1.308 ◆in the build CLAUDE_CODE_ALDER_WICKET ◇ Boolean. When true, the system prompt includes a # Delivering work section (Deliver what the user asked for, at the scope they intended.). The section is also included when the tengu_alder_wicket key in cached client data is true or the tengu_alder_wicket flag is on.v2.1.217 ○gone CLAUDE_CODE_ALLOW_OTHER_MODELS ◇ Claude Code now recognises CLAUDE_CODE_MCP_SERVE_WORKSPACE_ROOT alongside its other variables for running as an MCP serverv2.1.302 → v2.1.308 ◆in the build CLAUDE_CODE_ALT_SCREEN_FULL_REPAINT Set to 1 to repaint the entire screen on every frame in fullscreen rendering instead of sending incremental updates. Use this if fullscreen mode shows stale or misplaced text fragments. Claude Code enables this automatically for background sessions and agent view on Windowsv2.1.144 → v2.1.308 ◆in the build CLAUDE_CODE_ALWAYS_ENABLE_EFFORT Set to 1 to send the effort parameter with every request, even when Claude Code does not recognize the model ID as effort-capable. Use this when routing through an LLM gateway or third-party provider that serves models under custom identifiers. Models that reject the effort parameter at the API, including Claude 3 models, Sonnet 4.0 and 4.5, Opus 4.0 and 4.1, and Haiku 4.5, are still excluded so requests do not failv2.1.138 or earlier → v2.1.308 ◆in the build CLAUDE_CODE_AMBER_ASTROLABE ◇ Boolean (1, true, yes or on). When true, it forces the amber_astrolabe model capability on, which adds the system prompt section that opens You are operating autonomously. and tells the model to carry on with reversible actions without asking. The section still needs the tengu_amber_sextant flag (default true). When unset or false, the capability comes from the model's capability data, and the same section is also added for models with the fable_5_mitigations capability (on by default for claude-mythos-5).v2.1.217 → v2.1.308 ◆in the build CLAUDE_CODE_API_BASE_URL ◇ Remote settings can now supply many more environment variables, including auth tokens and home directory paths.v2.1.138 or earlier → v2.1.308 ◆in the build CLAUDE_CODE_API_KEY_FILE_DESCRIPTOR ◇ Auth tokens can now be handed to subprocesses over file descriptors instead of plaintext environment variables.v2.1.138 or earlier → v2.1.308 ◆in the build CLAUDE_CODE_API_KEY_HELPER_TTL_MS Interval in milliseconds at which credentials should be refreshed (when using apiKeyHelper)v2.1.138 or earlier → v2.1.308 ◆in the build CLAUDE_CODE_APPEND_PROMPT_HEAD ◇ In cloud sessions, CLAUDE_CODE_APPEND_PROMPT_HEAD can move a verified opening of --append-system-prompt text into the cacheable system promptv2.1.284 → v2.1.308 ◆in the build CLAUDE_CODE_ARG_KEY_SHAPE ◇ A new auth-adjacent variable name is listed but nothing in this build reads it.never mined ·never mined CLAUDE_CODE_ARTIFACT ◇ PR review artifact pages can show a pinned button that approves the pull request from your GitHub account.v2.1.166 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_ASSET_BASE_URL ◇ Three artifact environment variables are now recognised, with the API token stripped from spawned processes as a secret.v2.1.221 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_ASSETS ◇ Claude can now upload, list, read and delete images, video, PDFs and fonts attached to your published artifacts.v2.1.234 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_AUTO_OPEN Set to 0 to stop Claude Code from opening the browser automatically when a new artifact is publishedv2.1.172 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_COMMENT_FAST_ACK ◇ Claude can post a quick 'on it' note in artifact comment threads before the real reply, off unless enabled.v2.1.232 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_COMMENT_FAST_ACK_FIXED ◇ A second fast-ack mode lets Claude pick a fixed numbered acknowledgment instead of writing one, off behind two gatesv2.1.235 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_COMMENT_RESPONDER ◇ A read-only subagent that analyses one artifact comment thread exists but stays off.v2.1.227 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_COMMENTS Set to 0 to stop Claude reading and replying to comments on an artifact. Has no effect when CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC has turned artifacts off. Requires Claude Code v2.1.221 or laterv2.1.221 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_COMMENTS_AUTOREACT Set to 0 to stop Claude replying on its own to comments sent to it. Requires Claude Code v2.1.228 or laterv2.1.221 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_DB ◇ Published artifacts can gain a document database Claude reads and writes, off unless you set the env var.v2.1.224 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_DB_STR_REPLACE ◇ Artifact edits can now require a version match before writing, gated off by defaultv2.1.265 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_DELETE ◇ Claude can now read back a published artifact's contents; browsing, multi-file publishing and deletion are built but off.v2.1.239 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_DIRECT_UPLOAD ◇ Artifact availability now hinges on one gate, and direct upload kicks in for create paths.v2.1.172 → v2.1.238 ○gone CLAUDE_CODE_ARTIFACT_FRESH_READ ◇ Artifact reads for type-locked artifacts can now return data without first writing it to diskv2.1.274 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_HOT ◇ Artifact hot-reload now has its own dedicated env var and gate, separate from the artifact-types featurev2.1.267 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_LIVE_BASE_URL ◇ Three artifact environment variables are now recognised, with the API token stripped from spawned processes as a secret.v2.1.221 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_MCP ◇ No description from anybody.v2.1.172 → v2.1.173 ○gone CLAUDE_CODE_ARTIFACT_MULTI_FILE ◇ A new env var can switch on multi-file artifact support without waiting for the feature flag.v2.1.257 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_OPEN_ACTION ◇ Whether an artifact auto-opens in a local browser is now behind a new env var, off by default.v2.1.257 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_OPENING_PREFETCH ◇ Off switch for the artifact opening prefetch: only an explicit false value (0, false, no or off) has an effect, and it turns the prefetch off. The prefetch runs only when CLAUDE_CODE_ENTRYPOINT is remote_cowork and the artifact REPL tool or quickstart is on. Before the first assistant reply, if the opening user message names a newly created empty artifact, Claude Code reads that artifact type's instructions file ahead and attaches it to the turn as context.v2.1.274 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_PATH_PIN ◇ Boolean, default true when unset; 0, false, no or off turns it off. It controls artifact path pins: Claude Code records the SHA-256 of each artifact file Claude reads, lists or publishes, and refuses a publish that would replace or remove a file changed since Claude last saw it, or an already published file Claude has not read. The setting is captured when the session's artifact state is set up.v2.1.271 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_PIN ◇ The artifact publish action can offer a pin field, but it is refused outright in cloud sessions and otherwise depends on an env var or gate.v2.1.259 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_PRESENCE ◇ Artifacts can show who else is viewing live, but only if you set an environment variable.v2.1.251 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_PREVIEW ◇ An artifact preview action would screenshot a page locally at several widths, but it cannot run here.v2.1.247 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_PREVIEW_EMULATOR ◇ An emulator engine for artifact previews in remote sessions is built, but an internal switch that starts off keeps local Chrome in usev2.1.292 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_PUBLISH_EDITS ◇ The artifact publish tool can take a list of find-and-replace edits, applied to the file and saved, before it publishesv2.1.299 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_QUICKSTART ◇ A new artifact 'quickstart' feature exists in code but is off behind three separate switchesv2.1.271 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_REPL ◇ CLAUDE_CODE_ARTIFACT_REPL env var exists and is passed to subprocesses but nothing reads itv2.1.269 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_ROOM ◇ 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 CLAUDE_CODE_ARTIFACT_SHARE ◇ With CLAUDE_CODE_ARTIFACT_SHARE set, Claude can share an artifact with your organisation or named colleagues after you confirmv2.1.286 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_START_KIT ◇ A new gated 'start kit' feature can auto-bundle an artifact type, design system, and REPL toolv2.1.274 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_SYNC_BASE_URL ◇ You can point artifact syncing at a different host with a new environment variable.v2.1.238 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_TEXT_VARIANT ◇ The Artifact tool's instructions now come in versions v0, v1 and v2, picked once per session by an env var, the session origin or a flagv2.1.281 → v2.1.295 ○gone CLAUDE_CODE_ARTIFACT_TOOLSET ◇ Boolean (1/true/yes/on or 0/false/no/off) that turns the split artifact toolset on or off; when unset, the tengu_cobalt_plinth_damson flag decides (default false). The choice is made once per session. When on, and where the split tools are available, the Artifact tool's database actions belong to an ArtifactData tool, its watch and comment actions to ArtifactComments, and its verify and preview actions to ArtifactCheck, and Claude Code's instructions name those tools.v2.1.257 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_TYPE_CATALOG ◇ Claude could browse published artifact types to start from, but both actions are switched off.v2.1.242 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_TYPE_CLOUD_CREATE ◇ Listing and describing artifact types no longer prompts, and listing takes a filter.v2.1.242 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_TYPES ◇ You could start a private artifact from a published template, but two server switches keep it shut here.v2.1.236 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_VERIFY ◇ Claude could read console errors and failed loads from real viewers of an artifact, but it's switched off.v2.1.238 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_VERSIONS ◇ Claude Code now reads CLAUDE_CODE_MANAGED_CONFIG_PREFETCH at startup and keeps CLAUDE_CODE_ARTIFACT_VERSIONS out of programs it startsv2.1.292 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACT_VIEWER_BASE_URL ◇ An env var for a custom artifact viewer host is recognised but currently ignored.v2.1.229 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACTS_API_BASE_URL ◇ No description from anybody.v2.1.181 → v2.1.308 ◆in the build CLAUDE_CODE_ARTIFACTS_API_TOKEN ◇ Three artifact environment variables are now recognised, with the API token stripped from spawned processes as a secret.v2.1.221 → v2.1.308 ◆in the build CLAUDE_CODE_ATTACHMENTS_BY_PATH ◇ Setting CLAUDE_CODE_ATTACHMENTS_BY_PATH stops Claude Code uploading files it sends you, enables the send-file tool and keeps its results in subagentsv2.1.300 → v2.1.308 ◆in the build CLAUDE_CODE_ATTACHMENTS_IN_USER_DATA ◇ Remote sessions can keep attachments in a user-data folder when CLAUDE_CODE_ATTACHMENTS_IN_USER_DATA is set, unless data retention rules applyv2.1.301 → v2.1.308 ◆in the build CLAUDE_CODE_ATTRIBUTION_ANNOUNCEMENT ◇ Boolean (1/true/yes/on or 0/false/no/off); when unset, the tengu_reflective_wolf flag decides (default true). The value is read once and cached. When on, the Bash tool's instructions no longer contain the commit and PR attribution lines and say to end commit messages and PR bodies with "the attribution lines given in the conversation's system-reminder, when one is present", and the lines are sent in a system-reminder; when off, the lines are written into the Bash instructions directly.v2.1.268 → v2.1.308 ◆in the build CLAUDE_CODE_ATTRIBUTION_HEADER Set to 0 to omit the attribution block, which carries the client version and a prompt fingerprint, from the start of the system prompt. Caching on a direct connection to the Anthropic API is unaffected either way. In some direct-connection setups, Claude Code keeps the block on auto mode classifier requests even when you set 0. In System prompt attribution block, check which connections and credentials this covers. Before v2.1.181 the block included a per-request token on custom base URLs and Microsoft Foundry connections, so on those versions set it to 0 when your LLM gateway caches on the request body or forwards requests to a third-party provider, or when you connect to Microsoft Foundry directlyv2.1.138 or earlier → v2.1.308 ◆in the build CLAUDE_CODE_ATTRIBUTION_PROJECT ◇ Claude Code now recognises CLAUDE_CODE_ATTRIBUTION_PROJECT and drops a value that is not a valid claude.ai Project ID from saved jobsv2.1.304 → v2.1.308 ◆in the build CLAUDE_CODE_ATTRIBUTION_STATUS_TIMEOUT_MS ◇ Four more environment variables are now recognized and passed through by Claude Codev2.1.268 → v2.1.308 ◆in the build CLAUDE_CODE_AUTH_FAIL_EXIT_MS ◇ Integer milliseconds (minimum 0), default 600000 (10 minutes). It applies only when CLAUDE_CODE_REMOTE_SESSION_ID is set. When an OAuth 401 cannot be recovered and there is no refresh token, Claude Code notes the time of the first such failure; at a later failure at least this long after it, it logs that it is exiting so the runner can restart the session with fresh credentials, and exits with status 1 two seconds later. A successful recovery resets the clock, and 0 turns the exit off.v2.1.163 → v2.1.308 ◆in the build CLAUDE_CODE_AUTO_BACKGROUND_WORKER_CHECKIN_SECONDS When CLAUDE_AUTO_BACKGROUND_TASKS is enabled, seconds between reminders to Claude to check on background subagents that are still running. Accepts a plain integer from 1 to 86400 only; any other value or spelling reads as unset. When unset, there are no check-in reminders. Requires Claude Code v2.1.248 or laterv2.1.248 → v2.1.282 ○gone CLAUDE_CODE_AUTO_COMPACT_WINDOW Set the auto-compact window in tokens, from 100000 to 1000000. Accepts a plain integer such as 500000 only: a value like 500k reads as 500 and clamps to the 100K minimum. The effective window is also capped at the model's context window. Takes precedence over the /autocompact command, the --autocompact flag, and the autoCompactWindow setting. The status line's used_percentage always measures against the model's full context window, so once this variable is set, that percentage no longer indicates when compaction will runv2.1.138 or earlier → v2.1.308 ◆in the build CLAUDE_CODE_AUTO_CONNECT_IDE Override automatic IDE connection. By default, Claude Code connects automatically when launched inside a supported IDE's integrated terminal. Set to false to prevent this. Set to true to force a connection attempt when auto-detection fails, such as when tmux obscures the parent terminal. Takes precedence over the autoConnectIde global config settingv2.1.138 or earlier → v2.1.308 ◆in the build CLAUDE_CODE_AUTO_MODE_ARTIFACT_CONSENT_RULE ◇ The rule stopping Claude publishing artifacts you didn't ask for can now be removed remotely.never mined ·never mined