# Telemetry Report -- Claude Code v2.1.143

> Source: <https://gist.github.com/data2json/43611233bb1e6f9427e198d4cab4f4f0>
> Published: 2026-06-20 02:24:30+00:00

Telemetry Report -- Claude Code v2.1.143
Source: `deobfuscated_clean.js`

(14.5MB, 19,592 lines, Bun CJS bundle)
Date: 2026-06-20

`https://api.anthropic.com/api/claude_code/metrics`

(HTTP POST, JSON, 5s timeout)

Periodic: Every 5 minutes via `PeriodicExportingMetricReader`

On shutdown: `forceFlush()`

before process exit (2s timeout)
Startup perf: Single event at startup (or late startup), sampled at 0.5%
The metrics reader is added to the MeterProvider during app startup in `Bx_()`

. It is always added for first-party deployments
unless blocked by env vars. The `CLAUDE_CODE_ENABLE_TELEMETRY`

env var only controls the event logger (logs/traces), not the metric
s reader.

```
$V7() = !Wq$() && (PmH() || (Hq() && subscription in ["enterprise", "team"]))
```

Condition
Effect
`tYq() !== "default"`

Metrics disabled
Not first-party + not enterprise/team
Metrics disabled
First-party + not sandboxed + trusted
Metrics enabled
First-party + trusted + enterprise/team
Metrics enabled

Variable
Effect
Values
`CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC`

Disables metrics
Any truthy value
`DISABLE_TELEMETRY`

Disables metrics
Any truthy value
`DO_NOT_TRACK`

Disables metrics
`"1"`

, `"true"`

, `"yes"`

, `"on"`

`CLAUDE_CODE_ENABLE_TELEMETRY`

Controls event logger only (NOT metrics)
`"1"`

, `"true"`

, `"yes"`

, `"on"`

`CLAUDE_CODE_PROFILE_STARTUP`

Forces startup perf sampling
Any truthy value
`CLAUDE_CODE_OTEL_SHUTDOWN_TIMEOUT_MS`

Shutdown flush timeout
Number (default: 2000ms)
`OTEL_LOGS_EXPORT_INTERVAL`

Log export interval
Number (default: SDK default)
`OTEL_EXPORTER_OTLP_METRICS_PROTOCOL`

Metrics transport
`"http/json"`

or `"http/protobuf"`

Trust Gates (per-export check)
Before each export, the following must all pass:

Trust established (`Hq()`

-- sandboxed or authenticated)
Not in a restricted state (`!yT()`

)
Org settings don't disable it (`uJ7().enabled`

) -- checked every 24 hours
Resource attributes:

`service.name`

= "claude-code"
`service.version`

= "2.1.143"
`os.type`

(e.g., "linux", "darwin", "win32")
`os.version`

`host.arch`

(e.g., "x64", "arm64")
`wsl.version`

(if running under WSL)
`user.customer_type`

= "claude_ai" or "api"
`user.subscription_type`

= "free", "plus", "team", "enterprise", etc.
Metric data points:

`name`

-- event name (e.g., "tengu_git_operation")
`description`

-- human-readable description
`unit`

-- measurement unit
`value`

-- numeric value
`attributes`

-- key-value pairs from the event
`timestamp`

-- ISO 8601 format
Telemetry Events: 954 unique, 1,330 total calls
All event names are prefixed with `tengu_`

. Events fire on specific actions, not on a timer (except the periodic export of accumula
ted data).

API (16 events, 17 calls)
Event
Data
`tengu_api_529_background_dropped`

Background request dropped due to 529
`tengu_api_after_normalize`

Post-normalization state
`tengu_api_before_normalize`

Pre-normalization state
`tengu_api_cache_breakpoints`

Cache breakpoint info
`tengu_api_custom_529_overloaded_error`

529 overloaded response
`tengu_api_error`

API error details
`tengu_api_key_keychain_error`

Keychain access failure
`tengu_api_key_saved_to_config`

API key persisted to config
`tengu_api_key_saved_to_keychain`

API key persisted to keychain
`tengu_api_opus_fallback_triggered`

Opus fallback activated
`tengu_api_persistent_retry_wait`

Long retry wait
`tengu_api_query`

API request metadata
`tengu_api_retry`

Retry attempt
`tengu_api_retry_after_too_long`

Retry-After header exceeded
`tengu_api_slow_first_byte`

Slow TTFB
`tengu_api_success`

Successful API response (x2)

OAuth (45 events, 57 calls)
Event
Data
`tengu_oauth_401_recovered_from_disk`

Token recovered from disk
`tengu_oauth_401_recovered_from_keychain`

Token recovered from keychain
`tengu_oauth_401_sdk_callback_refreshed`

SDK callback refresh
`tengu_oauth_api_key`

API key auth success/failure, status code
`tengu_oauth_auth_code_received`

OAuth auth code
`tengu_oauth_automatic_redirect`

Auto-redirect (x2)
`tengu_oauth_automatic_redirect_error`

Redirect error
`tengu_oauth_bedrock_wizard_launched`

Bedrock auth wizard
`tengu_oauth_claudeai_forced`

Forced Claude.ai auth
`tengu_oauth_claudeai_selected`

Claude.ai selected
`tengu_oauth_env_var_api_key`

Env var API key detected
`tengu_oauth_fallback_to_api_key`

Fallback to API key
`tengu_oauth_foundry_wizard_launched`

Foundry auth wizard
`tengu_oauth_github_oauth_flow`

GitHub OAuth
`tengu_oauth_iam_role_prompted`

IAM role prompt
`tengu_oauth_keychain_error`

Keychain error
`tengu_oauth_keychain_success`

Keychain success
`tengu_oauth_login_success`

Login success
`tengu_oauth_mfa_required`

MFA required
`tengu_oauth_org_selection`

Org selection
`tengu_oauth_org_sso`

SSO org
`tengu_oauth_profile_fetch_success`

Profile fetch
`tengu_oauth_roles_stored`

`org_role`

`tengu_oauth_sso_login`

SSO login
`tengu_oauth_token_exchange_success`

Token exchange
`tengu_oauth_token_refresh_failure`

`error`

`tengu_oauth_token_refresh_success`

Token refresh
... and 19 more
Bedrock, Foundry, Vertex, WIF flows

MCP (43 events, 65 calls)
Event
Data
`tengu_mcp_add`

MCP config added (x3)
`tengu_mcp_auth_config_authenticate`

Auth config
`tengu_mcp_auth_config_clear`

Auth cleared
`tengu_mcp_channel_enable`

Channel enabled
`tengu_mcp_channel_flags`

Channel flags
`tengu_mcp_channel_gate`

Channel gate
`tengu_mcp_channel_message`

Channel message (x3)
`tengu_mcp_claudeai_proxy_401`

Proxy 401 (x2)
`tengu_mcp_degraded`

MCP degraded (x2)
`tengu_mcp_delete`

MCP config deleted (x2)
... and 33 more
Config load, transport, tool calls, errors

ToolUse (23 events, 29 calls)
Event
Data
`tengu_tool_empty_result`

Empty tool result
`tengu_tool_input_json_parse_fail`

JSON parse failure
`tengu_tool_result_pairing_repaired`

Result pairing fixed
`tengu_tool_result_persisted`

Result persisted
`tengu_tool_result_persisted_message_budget`

Budget exceeded
`tengu_tool_search_mcp_wait`

MCP search wait
`tengu_tool_search_mode_decision`

Search mode decision
`tengu_tool_search_outcome`

Search result
`tengu_tool_use_can_use_tool_allowed`

Tool allowed
`tengu_tool_use_can_use_tool_rejected`

Tool rejected
... and 13 more
Tool execution, timeouts, errors

File (21 events, 33 calls)
Event
Data
`tengu_file_changed`

File changed
`tengu_file_history_backup_deleted_file`

Backup of deleted file
`tengu_file_history_backup_file_created`

Backup file created
`tengu_file_history_backup_file_failed`

Backup failed
`tengu_file_history_resume_copy_failed`

Resume copy failed
`tengu_file_history_rewind_failed`

Rewind failed (x2)
`tengu_file_history_rewind_restore_file_failed`

Restore failed (x5)
`tengu_file_history_rewind_success`

Rewind success
`tengu_file_history_snapshot_failed`

Snapshot failed
`tengu_file_history_snapshot_success`

Snapshot success
... and 11 more
File reads, writes, changes, watch events

Session (14 events, 39 calls)
Event
Data
`tengu_session_all_projects_toggled`

All projects toggle
`tengu_session_branch_filter_toggled`

Branch filter toggle
`tengu_session_file_read`

File read event
`tengu_session_forked_branches_fetched`

Forked branches
`tengu_session_group_expanded`

Group expanded
`tengu_session_linked_to_pr`

PR linking
`tengu_session_persistence_failed`

Persistence failed (x2)
`tengu_session_preview_opened`

Preview opened
`tengu_session_rename_started`

Rename started
`tengu_session_renamed`

Rename completed
... and 4 more
Session switch, close, new, load

Agent (15 events, 21 calls)
Event
Data
`tengu_agent_color_set`

Agent color
`tengu_agent_created`

Agent created
`tengu_agent_definition_generated`

Definition generated
`tengu_agent_flag`

Agent flag
`tengu_agent_memory_loaded`

Memory loaded (x3)
`tengu_agent_name_set`

Agent name
`tengu_agent_parse_error`

Definition parse error
`tengu_agent_stop_hook_blocking`

Stop hook blocking
`tengu_agent_stop_hook_error`

Stop hook error (x2)
`tengu_agent_stop_hook_max_turns`

Max turns reached
`tengu_agent_stop_hook_success`

Stop hook success
`tengu_agent_summary_skipped`

Summary skipped
`tengu_agent_tool_completed`

Tool completed
`tengu_agent_tool_selected`

Tool selected
`tengu_agent_tool_terminated`

Tool terminated (x4)

Plugin (26 events, 28 calls)
Event
Data
`tengu_plugin_command_failed`

Plugin command failed
`tengu_plugin_details_command`

Plugin details
`tengu_plugin_disable_command`

Disable command (x2)
`tengu_plugin_disabled_all_cli`

Disabled all
`tengu_plugin_disabled_cli`

Disabled single
`tengu_plugin_enable_command`

Enable command
`tengu_plugin_enabled_cli`

Enabled single
`tengu_plugin_enabled_for_session`

Session enable
`tengu_plugin_folder_shadowed`

Folder shadowed
`tengu_plugin_hint_detected`

Hint detected
... and 16 more
Install, load, error, config, reload

Config (7 events, 18 calls)
Event
Data
`tengu_config_auth_loss_prevented`

Auth loss prevented (x4)
`tengu_config_cache_stats`

Cache stats
`tengu_config_changed`

Config changed (x9)
`tengu_config_lock_contention`

Lock contention
`tengu_config_model_changed`

Model changed
`tengu_config_parse_error`

Parse error
`tengu_config_stale_write`

Stale write

Shell (5 events, 6 calls)
Event
Data
`tengu_shell_completion_failed`

Command failed
`tengu_shell_set_cwd`

Working directory changed (x2)
`tengu_shell_snapshot_error`

Snapshot error
`tengu_shell_snapshot_failed`

Snapshot failed
`tengu_shell_unknown_error`

Unknown error

Permission (2 events, 5 calls)
Event
Data
`tengu_permission_request_escape`

Escape attempt (x3)
`tengu_permission_request_option_selected`

User's choice

Model (8 events, 10 calls)
Event
Data
`tengu_model_command_inline`

Inline model change (x2)
`tengu_model_command_inline_help`

Inline help (x1)
`tengu_model_command_menu`

Menu model change (x2)
`tengu_model_command_menu_effort`

Effort setting
`tengu_model_fallback_triggered`

Model fallback
`tengu_model_picker_hotkey`

Hotkey picker
`tengu_model_response_keyword_detected`

Keyword in response
`tengu_model_whitespace_response`

Whitespace-only response

Input (5 events, 9 calls)
Event
Data
`tengu_input_bash`

Bash input
`tengu_input_command`

Command input (x2)
`tengu_input_prompt`

Prompt input (x2)
`tengu_input_slash_invalid`

Invalid slash command (x3)
`tengu_input_slash_missing`

Missing slash

Daemon (10 events, 11 calls)
Event
Data
`tengu_daemon_config_reload`

Config reloaded
`tengu_daemon_idle_exit`

Idle exit
`tengu_daemon_lease`

Lease (x2)
`tengu_daemon_peer_uid_reject`

UID rejected
`tengu_daemon_self_restart_on_upgrade`

Self-restart
`tengu_daemon_start`

Daemon start
`tengu_daemon_worker_crash`

Worker crash
`tengu_daemon_worker_permanent_exit`

Permanent exit
`tengu_daemon_yield`

Yield
`tengu_daemon_yield_takeover`

Yield takeover

FeatureFlags (3 events, 3 calls)
Event
Data
`tengu_feature_bad`

`feature_name`

, `error_code`

, additional data
`tengu_feature_ok`

`feature_name`

`tengu_feature_sad`

`feature_name`

, `error_code`

FastMode (4 events, 5 calls)
Event
Data
`tengu_fast_mode_fallback_triggered`

`cooldown_duration_ms`

, `cooldown_reason`

`tengu_fast_mode_overage_rejected`

`overage_disabled_reason`

`tengu_fast_mode_picker_shown`

Picker displayed
`tengu_fast_mode_toggled`

Toggle (x2)

Feedback (3 events, 12 calls)
Event
Data
`tengu_accept_feedback_mode_collapsed`

Mode collapsed (x4)
`tengu_accept_feedback_mode_entered`

Mode entered (x4)
`tengu_accept_submitted`

Feedback submitted (x4)

AskUserQuestion (4 events, 4 calls)
Event
Data
`tengu_ask_user_question_accepted`

Question accepted
`tengu_ask_user_question_finish_plan_interview`

Plan interview finished
`tengu_ask_user_question_rejected`

Question rejected
`tengu_ask_user_question_respond_to_claude`

Responded to Claude

Event
Data
`tengu_git_index_lock_error`

Lock error
`tengu_git_operation`

Git operation (x7)

Prompt (4 events, 12 calls)
Event
Data
`tengu_prompt_cache_break`

Cache broken
`tengu_prompt_cache_diagnosis_received`

Diagnosis received
`tengu_prompt_suggestion`

Suggestion (x4)
`tengu_prompt_suggestion_init`

Suggestion init (x6)

Event
Data
`tengu_rate_limit_lever_hint`

Rate limit hint
`tengu_rate_limit_options_menu_cancel`

Cancelled
`tengu_rate_limit_options_menu_select_extra_usage`

Extra usage
`tengu_rate_limit_options_menu_select_team`

Team plan
`tengu_rate_limit_options_menu_select_upgrade`

Upgrade

Startup (3 events, 3 calls)
Event
Data
`tengu_startup_manual_model_config`

Manual model config
`tengu_startup_perf`

`import_time`

, `main_imports`

, `init_time`

, `settings_time`

, `tools_loaded`

, `mcp_configs`

, `plugins_init`

, `headless_setup`

, `force_login_org`

, `mcp_connect`

, `mcp_connect_user`

, `mcp_connect_connector`

, `growthbook_init`

, `prewait`

, `s
andbox_init`, `

process_user_input`, `

total_time`, `

gap_unaccounted_ms`, `

free_mem_mb`, `

load_avg_1m`, `

checkpoint_count`
`tengu_startup_telemetry`

Telemetry init

Event
Data
`tengu_exit_background_work_prompt`

Background work prompt
`tengu_exit_plan_mode_called_outside_plan`

Plan mode outside plan
`tengu_unclean_exit`

`session_age_sec`

Event
Data
`tengu_wif_user_oauth_lock_acquired`

Lock acquired
`tengu_wif_user_oauth_lock_released`

Lock released
`tengu_wif_user_oauth_lock_retry`

Lock retry
`tengu_wif_user_oauth_lock_retry_limit`

Retry limit

Category
Events
What It Tracks
Other
632
@-mentions, attachments, cursor position, rename, rm, slow ops, code index, diff, edit, help, login, memory, op
en, skill, stream, update, write, bug reports, claude chrome, config, headless, input, intent, json, keychain, kill, lifecycle, mod
el, multi, npm, ollama, parse, rule, search, shell, skill, token, tool_use, user_input, zod, and 600+ more granular events
Advisor
6
Advisor dialog, commands, tool calls/interrupts/tokens
Agentic
1
Agentic search cancelled
Ant
1
Overly broad bash detected
ApiKeyHelper
1
Missing trust11
Assistant
1
Assistant install
Async
1
Agent stall timeout
BugReport
2
Bug report submitted/failed
CLI
1
CLI flags
Claude
11
Chrome onboarding, CLAUDE.md dialogs, skill loaded, install command
Code
2
Code indexing, prompt ignored
Diff
1
Diff tool changed
Edit
1
Edit string lengths
Headless
3
Latency, MCP prewait, plugin install
Help
1
Help toggled
Login
1
Login from refresh token
Memory
2
Threshold crossed, toggled
Org
1
Penguin mode fetch failed
Open
1
Open agents via left panel
Skill
4
File changed, loaded, tool invocation
Stream
2
Loop exited, no events
Update
2
Update check, refused

OS type and version
Host architecture (x64, arm64, etc.)
WSL version (if applicable)
Free memory (startup perf only)
Load average 1m (startup perf only)
Customer type: "claude_ai" or "api"
Subscription type: "free", "plus", "team", "enterprise", etc.
Service version: "2.1.143"
Every tool call (allowed, rejected, terminated, empty result, JSON parse fail)
Every file operation (read, write, rename, rm, watch, history snapshot/rewind)
Every shell command (completion, failure, CWD change, snapshot)
Every permission prompt (escape attempt, option selected)
Every model change (inline, menu, picker, fallback, whitespace response)
Every input type (bash, command, prompt, invalid/missing slash)
Every @-mention parsing result (success, error, directory, filename, MCP resource)
Every attachment (size, too-large, compute duration)
Every cursor position change
Every session event (file read, branch ops, persistence, rename, preview, PR linking)
Every plugin action (enable, disable, install, load, error, config)
Every MCP action (config add/delete, auth, channel, proxy 401, degraded)
Every git operation and lock error
Every config change, cache stat, lock contention, parse error
Every daemon event (start, idle exit, lease, crash, restart)
Every feedback interaction (mode entered/collapsed, submitted)
Every AskUserQuestion interaction (accepted, rejected, plan interview)
Every rate limit interaction (hint, options menu)
Every fast mode interaction (toggled, picker, fallback, overage)
Every feature flag outcome (ok, bad, sad)
Every agent lifecycle event (creation, memory, stop hooks, tool events)
Every advisor interaction (dialog, command, tool call)
Every OAuth event (token refresh, 401 recovery, auth code, SSO, MFA)
Every API event (query, error, retry, 529, slow first byte, opus fallback)
Every prompt suggestion (init, display)
Every prompt cache event (break, diagnosis)
Every stream event (loop exited, no events)
Every update check and refusal
Every bug report submission
Every skill event (load, file changed, tool invocation)
Every Ollama interaction
Every NPM interaction
Every intent detection
Every JSON parse event
Every keychain event
Every lifecycle event
Every rule event
Every search event
Every token event
Every zod validation event
Startup Performance (sampled 0.5%)
Per-stage timing: import, main imports, init, settings, tools, MCP configs, plugins, headless setup, force login org, MCP connect
, MCP connect user, MCP connect connector, GrowthBook init, prewait, sandbox init, process user input
Total time
Unaccounted time gap
Free memory (MB)
Load average (1m)
Checkpoint count
Message content -- no prompts, responses, or conversation text
File contents -- no file data, only operation metadata
Shell command output -- no stdout/stderr from commands
API keys or tokens -- only success/failure events, never the values
PII -- no names, emails, addresses, or personal identifiers
Source code -- no code content, only operation counts
Terminal content -- no terminal output or screen content
Clipboard data -- no clipboard access
Network request bodies -- only event metadata, not request/response payloads
Environment variable values -- only presence/absence of specific vars, not their values
Aspect
Value
Unique event names
954
Total event calls in code
1,330
Export interval
5 minutes
Startup perf sampling
0.5%
Default enabled
Yes (first-party deployments)
Disable method
`DISABLE_TELEMETRY=1`

or `DO_NOT_TRACK=1`

Data destination
`api.anthropic.com/api/claude_code/metrics`

Transport
HTTP POST, JSON
Auth
Bearer token in request headers
PII collected
None
Message content collected
None
File content collected
None
Command output collected
None