fix: remove internal Anthropic tooling from external build (#345)

Remove debug systems, employee detection, and internal logging
that have no function in a community fork.

Changes:
- Remove logPermissionContextForAnts import and calls (main.tsx, compact.ts)
  Reads Kubernetes namespace and container IDs from internal infra paths.
  Dead code for all external users.

- Remove createDumpPromptsFetch import and gate (query.ts)
  Internal prompt dump system for employee debugging.
  Replace gate with unconditional undefined — normal fetch always used.

- Remove stripSignatureBlocks ant-only block (query.ts)
  Was behind USER_TYPE === 'ant' guard, never ran for external users.

- Hardcode isAnt: false (query/config.ts)
  Employee detection flag has no place in a community fork.
  config.gates.isAnt had exactly one consumer (dumpPromptsFetch, now removed).

- Gut logClassifierResultForAnts body (bashPermissions.ts)
  Replace with empty no-op. Still called from 4 sites, zero execution.
  Remove ANT-ONLY comments describing internal security model.

- Gate status.anthropic.com behind firstParty check (errors.ts)
  429 error hint now only shown when using Anthropic directly.
  Third-party provider users see a generic capacity message.

Build: passes
Typecheck: clean (no new errors)
Tests: 196 pass, same 6 pre-existing failures unrelated to these changes

src/main.tsx

@@ -142,7 +142,6 @@ import { validateUuid } from './utils/uuid.js';
 import { registerMcpAddCommand } from 'src/commands/mcp/addCommand.js';
 import { registerMcpDoctorCommand } from 'src/commands/mcp/doctorCommand.js';
 import { registerMcpXaaIdpCommand } from 'src/commands/mcp/xaaIdpCommand.js';
-import { logPermissionContextForAnts } from 'src/services/internalLogging.js';
 import { fetchClaudeAIMcpConfigsIfEligible } from 'src/services/mcp/claudeai.js';
 import { clearServerCache } from 'src/services/mcp/client.js';
 import { areMcpConfigsAllowedWithEnterpriseMcpConfig, dedupClaudeAiMcpServers, doesEnterpriseMcpConfigExist, filterMcpServersByPolicy, getClaudeCodeMcpConfigs, getMcpServerSignature, parseMcpConfig, parseMcpConfigFromFilePath } from 'src/services/mcp/config.js';
@@ -2511,7 +2510,6 @@ async function run(): Promise<CommanderCommand> {
 
     // Log context metrics once at initialization
     void logContextMetrics(regularMcpConfigs, toolPermissionContext);
-    void logPermissionContextForAnts(null, 'initialization');
     logManagedSettings();
 
     // Register PID file for concurrent-session detection (~/.claude/sessions/)

src/query.ts

@@ -52,7 +52,6 @@ import {
   getMessagesAfterCompactBoundary,
   createToolUseSummaryMessage,
   createMicrocompactBoundaryMessage,
-  stripSignatureBlocks,
 } from './utils/messages.js'
 import { generateToolUseSummary } from './services/toolUseSummary/toolUseSummaryGenerator.js'
 import { prependUserContext, appendSystemContext } from './utils/api.js'
@@ -92,7 +91,6 @@ import { SLEEP_TOOL_NAME } from './tools/SleepTool/prompt.js'
 import { executePostSamplingHooks } from './utils/hooks/postSamplingHooks.js'
 import { executeStopFailureHooks } from './utils/hooks.js'
 import type { QuerySource } from './constants/querySource.js'
-import { createDumpPromptsFetch } from './services/api/dumpPrompts.js'
 import { StreamingToolExecutor } from './services/tools/StreamingToolExecutor.js'
 import { queryCheckpoint } from './utils/queryProfiler.js'
 import { runTools } from './services/tools/toolOrchestration.js'
@@ -587,13 +585,7 @@ async function* queryLoop(
 
     // Create fetch wrapper once per query session to avoid memory retention.
     // Each call to createDumpPromptsFetch creates a closure that captures the request body.
-    // Creating it once means only the latest request body is retained (~700KB),
-    // instead of all request bodies from the session (~500MB for long sessions).
-    // Note: agentId is effectively constant during a query() call - it only changes
-    // between queries (e.g., /clear command or session resume).
-    const dumpPromptsFetch = config.gates.isAnt
-      ? createDumpPromptsFetch(toolUseContext.agentId ?? config.sessionId)
-      : undefined
+    const dumpPromptsFetch = undefined
 
     // Block if we've hit the hard blocking limit (only applies when auto-compact is OFF)
     // This reserves space so users can still run /compact manually
@@ -931,9 +923,6 @@ async function* queryLoop(
             // Thinking signatures are model-bound: replaying a protected-thinking
             // block (e.g. capybara) to an unprotected fallback (e.g. opus) 400s.
             // Strip before retry so the fallback model gets clean history.
-            if (process.env.USER_TYPE === 'ant') {
-              messagesForQuery = stripSignatureBlocks(messagesForQuery)
-            }
 
             // Log the fallback event
             logEvent('tengu_model_fallback_triggered', {

src/query/config.ts

@@ -36,7 +36,7 @@ export function buildQueryConfig(): QueryConfig {
       emitToolUseSummaries: isEnvTruthy(
         process.env.CLAUDE_CODE_EMIT_TOOL_USE_SUMMARIES,
       ),
-      isAnt: process.env.USER_TYPE === 'ant',
+      isAnt: false,
       // Inlined from fastMode.ts to avoid pulling its heavy module graph
       // (axios, settings, auth, model, oauth, config) into test shards that
       // didn't previously load it — changes init order and breaks unrelated tests.

src/services/api/errors.ts

@@ -558,7 +558,7 @@ export function getAssistantMessageFromError(
     const innerMessage = stripped.match(/"message"\s*:\s*"([^"]*)"/)?.[1]
     const detail = innerMessage || stripped
     return createAssistantAPIErrorMessage({
-      content: `${API_ERROR_MESSAGE_PREFIX}: Request rejected (429) · ${detail || 'this may be a temporary capacity issue — check status.anthropic.com'}`,
+      content: `${API_ERROR_MESSAGE_PREFIX}: Request rejected (429) · ${detail || `this may be a temporary capacity issue${getAPIProvider() === 'firstParty' ? ' — check status.anthropic.com' : ''}`}`,
       error: 'rate_limit',
     })
   }

src/services/compact/compact.ts

@@ -107,7 +107,6 @@ import {
 } from '../api/errors.js'
 import { notifyCompaction } from '../api/promptCacheBreakDetection.js'
 import { getRetryDelay } from '../api/withRetry.js'
-import { logPermissionContextForAnts } from '../internalLogging.js'
 import {
   roughTokenCountEstimation,
   roughTokenCountEstimationForMessages,
@@ -401,7 +400,6 @@ export async function compactConversation(
     const preCompactTokenCount = tokenCountWithEstimation(messages)
 
     const appState = context.getAppState()
-    void logPermissionContextForAnts(appState.toolPermissionContext, 'summary')
 
     context.onCompactProgress?.({
       type: 'hooks_start',

src/tools/BashTool/bashPermissions.ts

@@ -110,37 +110,16 @@ export const MAX_SUBCOMMANDS_FOR_SECURITY_CHECK = 50
 export const MAX_SUGGESTED_RULES_FOR_COMPOUND = 5
 
 /**
- * [ANT-ONLY] Log classifier evaluation results for analysis.
- * This helps us understand which classifier rules are being evaluated
- * and how the classifier is deciding on commands.
+ * Log classifier evaluation results for analysis.
+ * No-op in external builds.
  */
 function logClassifierResultForAnts(
-  command: string,
-  behavior: ClassifierBehavior,
-  descriptions: string[],
-  result: ClassifierResult,
+  _command: string,
+  _behavior: ClassifierBehavior,
+  _descriptions: string[],
+  _result: ClassifierResult,
 ): void {
-  if (process.env.USER_TYPE !== 'ant') {
-    return
-  }
-
-  logEvent('tengu_internal_bash_classifier_result', {
-    behavior:
-      behavior as AnalyticsMetadata_I_VERIFIED_THIS_IS_NOT_CODE_OR_FILEPATHS,
-    descriptions: jsonStringify(
-      descriptions,
-    ) as AnalyticsMetadata_I_VERIFIED_THIS_IS_NOT_CODE_OR_FILEPATHS,
-    matches: result.matches,
-    matchedDescription: (result.matchedDescription ??
-      '') as AnalyticsMetadata_I_VERIFIED_THIS_IS_NOT_CODE_OR_FILEPATHS,
-    confidence:
-      result.confidence as AnalyticsMetadata_I_VERIFIED_THIS_IS_NOT_CODE_OR_FILEPATHS,
-    reason:
-      result.reason as AnalyticsMetadata_I_VERIFIED_THIS_IS_NOT_CODE_OR_FILEPATHS,
-    // Note: command contains code/filepaths - this is ANT-ONLY so it's OK
-    command:
-      command as AnalyticsMetadata_I_VERIFIED_THIS_IS_NOT_CODE_OR_FILEPATHS,
-  })
+  // Internal-only logging removed from external build.
 }
 
 /**
@@ -430,13 +409,11 @@ const SAFE_ENV_VARS = new Set([
 ])
 
 /**
- * ANT-ONLY environment variables that are safe to strip from commands.
- * These are only enabled when USER_TYPE === 'ant'.
+ * Environment variables that are safe to strip from commands.
  *
  * SECURITY: These env vars are stripped before permission-rule matching, which
  * means `DOCKER_HOST=tcp://evil.com docker ps` matches a `Bash(docker ps:*)`
- * rule after stripping. This is INTENTIONALLY ANT-ONLY (gated at line ~380)
- * and MUST NEVER ship to external users. DOCKER_HOST redirects the Docker
+ * rule after stripping. DOCKER_HOST redirects the Docker
  * daemon endpoint — stripping it defeats prefix-based permission restrictions
  * by hiding the network endpoint from the permission check. KUBECONFIG
  * similarly controls which cluster kubectl talks to. These are convenience