kshitij 7b006ea6e8 feat(stt): idle unload for local whisper model ... The local faster-whisper model singleton (_local_model) is loaded once and never released — the 'base' model holds ~370 MB of RAM/VRAM for the entire lifetime of the process, even when no voice messages arrive for hours or days. On long-running gateway processes (especially with local LLMs competing for the same GPU) this is wasteful. Add a config-driven idle unload: after stt.local.unload_after_idle_seconds (default 0 = never) of no transcription activity, a lightweight daemon thread sets _local_model = None so the Python GC can reclaim the ctranslate2 objects. The next voice message reloads the model transparently (the existing lazy-load path handles it). The watcher: - Checks every 30s whether idle time exceeds the configured threshold - Acquires _local_model_lock before unloading (prevents races with concurrent transcriptions that are mid-load) - Exits immediately if the model is already None (unloaded by another path, e.g. the CUDA fallback eviction) - Is restarted by each transcription with the current config value, so changing stt.local.unload_after_idle_seconds in config.yaml takes effect on the next voice message without a process restart Default is 0 (never unload) — zero behavior change for existing users. Recommended value for gateway processes: 300 (5 minutes). 15 tests: config resolution (garbage/negative/None fallbacks), unload safety (already-None, lock acquisition), touch timestamp, watcher lifecycle (unload after timeout, no unload within timeout, exits when model already None, stopped on new start). Existing STT test suite unchanged.		2026-08-07 19:26:12 +05:30
..
acp	…
acp_adapter	…
agent	fix(redact): narrow control-split join guard to line-crossing spans	2026-08-07 17:50:03 +05:30
ci	fix(ci): follow artifact download redirect without auth	2026-08-05 11:16:18 -04:00
cli	…
computer_use	…
conformance	…
cron	test(cron): pin fail-closed TERMINAL_CWD lock timeout behavior	2026-08-07 18:11:41 +05:30
dashboard	…
docker	…
e2e	…
fakes	…
fixtures	…
gateway	fix(slack): dedupe thread-qualified channel lookups (#80668)	2026-08-07 18:40:33 +05:30
hermes_cli	refactor(cron): polish registration partial-failure surfaces	2026-08-07 17:45:06 +05:30
hermes_state	test: fold review findings — plain fixture, call-shaped probe guard, public-API row counting	2026-08-07 15:42:49 +05:30
honcho_plugin	…
install	test(install): prove updating from a release reaches this commit	2026-08-04 17:36:26 -04:00
integration	…
manual	…
monitoring	…
openviking_plugin	…
plugins	fix(cron): surface initial scheduler registration failures	2026-08-07 17:45:06 +05:30
providers	fix(cache): scope prompt_cache_key by session to stop cross-session bucket sharing	2026-08-05 12:42:46 +05:30
run_agent	fix(streaming): close the same mid-tool-call drop gap on the Anthropic path	2026-08-07 18:07:19 +05:30
scripts	…
secret_sources	…
skills	feat(skills): actual-setup optional skill + provider docs	2026-08-05 14:08:32 -07:00
state	fix: bind continuation-marker exclusions to the queried parent (fail-open fix)	2026-08-07 13:24:56 +05:30
stress	…
tools	feat(stt): idle unload for local whisper model	2026-08-07 19:26:12 +05:30
tui_gateway	fix(agent): the desktop's tools reach it on remote and cloud backends too	2026-08-06 19:35:47 -06:00
website	…
__init__.py	…
conftest.py	fix(tests): fail hard when pytest resolves the production state.db (live-DB isolation guard)	2026-08-06 07:49:50 -07:00
run_interrupt_test.py	…
test_account_usage.py	…
test_atomic_replace_symlinks.py	…
test_atomic_write_text_metadata.py	fix(utils): tighten create_mode semantics and close the yaml 0600 transit window	2026-08-06 05:00:17 +05:30
test_audio_playback_guard.py	…
test_background_review_list_shapes.py	…
test_background_review_session_isolation.py	…
test_base_url_hostname.py	…
test_batch_runner_checkpoint.py	…
test_batch_runner_durability.py	…
test_bitwarden_secrets.py	…
test_cli_manual_compress.py	…
test_cli_skin_integration.py	…
test_code_skew.py	…
test_command_secret_source.py	…
test_conftest_wal_gate.py	…
test_copilot_initiator.py	…
test_credential_file_permissions.py	…
test_ctx_halving_fix.py	…
test_delegate_cascade_49148.py	…
test_dispatch_session_id.py	…
test_empty_model_fallback.py	feat(models): add qwen3.8-max to Nous portal + OpenRouter catalogs, replacing qwen3.7-max	2026-08-03 17:19:49 -07:00
test_empty_session_hygiene.py	…
test_engines_satisfiable.py	…
test_env_loader_applied_homes.py	…
test_env_loader_op_bootstrap.py	…
test_env_loader_secret_sources.py	…
test_evidence_store.py	…
test_fast_safe_load.py	…
test_fts_cjk_bigram.py	…
test_fts_update_of_narrowing.py	…
test_gateway_streaming_nested_config.py	…
test_get_tool_definitions_cache_isolation.py	…
test_hermes_bootstrap.py	…
test_hermes_constants.py	…
test_hermes_home_profile_warning.py	…
test_hermes_logging.py	…
test_hermes_state.py	perf(fts): drain trash tables with a high-water marker instead of re-scanning	2026-08-07 18:40:47 +05:30
test_hermes_state_compression_busy_retry.py	…
test_hermes_state_compression_locks.py	…
test_hermes_state_readonly_preflight.py	…
test_hermes_state_wal_fallback.py	…
test_hermetic_side_effect_guards.py	…
test_honcho_client_concurrency.py	…
test_honcho_client_config.py	…
test_honcho_session_context.py	…
test_honcho_startup_fail_open.py	…
test_install_autostash_conflict_recovery.py	…
test_install_commit_pin_rollback.py	…
test_install_diverged_update.py	…
test_install_lockfile_churn.py	…
test_install_macos_launcher.py	…
test_install_no_initial_commit.py	…
test_install_ps1_ascii_only.py	…
test_install_ps1_native_stderr_eap.py	…
test_install_ps1_node_path_for_npm.py	…
test_install_ps1_python_fallback_venv.py	…
test_install_ps1_uv_powershell_host.py	…
test_install_ps1_web_server_syntax_probe.py	…
test_install_sh_acp_launcher.py	…
test_install_sh_bootstrap_marker.py	…
test_install_sh_browser_install.py	…
test_install_sh_install_method_stamp.py	…
test_install_sh_node_global_prefix.py	…
test_install_sh_pythonpath_sanitization.py	…
test_install_sh_root_fhs_uv_python_path.py	…
test_install_sh_setup_wizard_tty_probe.py	…
test_install_sh_symlink_stomp.py	…
test_install_sh_termux_network_prereqs.py	…
test_install_unmerged_index.py	…
test_ipv4_preference.py	…
test_iron_proxy.py	…
test_iron_proxy_cli.py	…
test_iron_proxy_e2e.py	…
test_journal_mode_config.py	…
test_lazy_session_regressions.py	…
test_live_system_guard.py	…
test_live_system_guard_self_test.py	…
test_log_isolation.py	…
test_managed_runtime_resolution.py	…
test_mcp_serve.py	…
test_message_reactions.py	…
test_mini_swe_runner.py	…
test_minimax_model_validation.py	…
test_minimax_oauth.py	…
test_minisweagent_path.py	…
test_model_forces_max_completion_tokens.py	…
test_model_picker_scroll.py	…
test_model_tools.py	…
test_model_tools_async_bridge.py	…
test_no_shadowed_test_definitions.py	…
test_ollama_num_ctx.py	…
test_onepassword_secrets.py	…
test_output_cap_parsing.py	…
test_packaging_build_guard.py	…
test_packaging_metadata.py	…
test_plugin_skills.py	…
test_plugin_utils.py	…
test_process_loop_event_loop_warning.py	…
test_profile_isolation_runtime.py	…
test_project_metadata.py	…
test_pty_keepalive_ws.py	…
test_pty_session.py	…
test_retry_utils.py	…
test_run_tests_parallel.py	…
test_run_tests_parallel_stdio.py	…
test_sanitize_tool_error.py	…
test_schema_read_probe.py	fix(dashboard): derive the stale-schema read probe from SCHEMA_SQL	2026-08-07 00:41:58 -04:00
test_search_slow_query_log.py	…
test_secret_scope_plugin_families.py	…
test_session_db_read_path_split.py	fix(state): route session-resume reads through the WAL read-only connection	2026-08-03 21:02:11 +05:30
test_session_skill_previews.py	…
test_session_system_prompt_dedup.py	…
test_session_vacuum_config.py	…
test_session_workspace_binding.py	…
test_slack_thread_require_mention.py	…
test_slash_worker_watchdog.py	…
test_sql_injection.py	…
test_sqlite_lock_safe_inspection.py	…
test_sqlite_wal_reset_gate.py	fix(db): never downgrade journal mode on a database with concurrent openers	2026-08-06 07:49:42 -07:00
test_stale_tool_call_marker_session_repair.py	fix(cli): back up state.db before clean-markers writes by default	2026-08-04 11:26:15 +05:30
test_stale_utils_module_import.py	…
test_state_db_malformed_repair.py	…
test_subprocess_home_isolation.py	…
test_telegram_polling_progress_ptb.py	…
test_termux_all_extra_compat.py	…
test_timezone.py	…
test_tini_shim.py	…
test_toolset_distributions.py	…
test_toolsets.py	fix(agent): the desktop's tools reach it on remote and cloud backends too	2026-08-06 19:35:47 -06:00
test_trajectory_compressor.py	…
test_trajectory_compressor_async.py	…
test_transform_llm_output_hook.py	…
test_transform_tool_result_hook.py	…
test_tui_entry_mcp_owner.py	…
test_tui_gateway_loop_noise.py	…
test_tui_gateway_queue_on_busy.py	…
test_tui_gateway_server.py	fix(gateway): make a history-dropping submit prove it meant to	2026-08-07 18:20:19 +05:30
test_tui_gateway_server_crash_history.py	fix: preserve session history when a turn crashes	2026-08-06 23:08:23 -05:00
test_tui_gateway_ws.py	…
test_tui_mcp_late_refresh.py	fix(agent): the desktop's tools reach it on remote and cloud backends too	2026-08-06 19:35:47 -06:00
test_utils_truthy_values.py	…
test_voice_max_recording_seconds.py	…
test_wal_checkpoint_strategy.py	…
test_web_server.py	…
test_web_server_sessiondb_eventloop.py	…
test_web_server_status_topology_cache.py	…
test_windows_subprocess_no_window_flags.py	…
test_yaml_indent_consistency_31999.py	…
test_yuanbao_integration.py	…
test_yuanbao_markdown.py	…
test_yuanbao_pipeline.py	…
test_yuanbao_proto.py	…
test_yuanbao_reconnect_set_active.py	…
test_yuanbao_shutdown.py	…
test_zeroed_state_db.py	…