llamastash
Zero-overhead, terminal-native local-LLM runtime manager. Launches, supervises, and routes local models behind one OpenAI-compatible endpoint.
File Explorer
Download Latest Version (.zip)- audit.toml
- pre-commit
- pre-push
- marketplace.json
- nextest.toml
- bug_report.md
- feature_request.md
- default.md
- release.md
- ci.yml
- publish-aur-manual.yml
- publish-cargo-manual.yml
- publish-homebrew-manual.yml
- regenerate-benchmark-snapshot.yml
- release.yml
- dependabot.yml
- 00-hero-all-hardware.png
- 00-hero-all-hardware.svg
- 01-amd-apu-decode.png
- 01-amd-apu-decode.svg
- 02-amd-apu-ttft.png
- 02-amd-apu-ttft.svg
- 03-m1-decode-defaults-vs-normalized.png
- 03-m1-decode-defaults-vs-normalized.svg
- 04-nvidia-decode-cuda-vs-vulkan.png
- 04-nvidia-decode-cuda-vs-vulkan.svg
- 05-amd-apu-vulkan-addendum.png
- 05-amd-apu-vulkan-addendum.svg
- 06-amd-apu-combined.png
- 06-amd-apu-combined.svg
- 07-amd-apu-combined-agent-decode.png
- 07-amd-apu-combined-agent-decode.svg
- 08-suite-a-overhead.png
- 08-suite-a-overhead.svg
- 09-suite-c-proxy.png
- 09-suite-c-proxy.svg
- demo.cast
- demo.raw.cast
- init.gif
- logo-h.jpg
- logo-h.png
- logo-h.webp
- logo-m.png
- logo-only-xs.png
- logo-only.png
- logo-s.png
- logo-xs.png
- logo-xss.png
- logo.png
- presets.gif
- tui.gif
- tui_1.png
- tui_2.png
- tui_3.png
- tui_4.png
- tui_5.png
- tui_6.png
- tui_7.png
- proxy_overhead.rs
- flowz13-arch-hip-20260519T202701Z.json
- benchmark-snapshot.json
- gguf-publisher-allowlist.yaml
- task-hints.yaml
- PKGBUILD.template
- PKGBUILD.template
- PKGBUILD
- packager.py
- packager_test.py
- llamastash.rb.template
- packager.py
- packager_test.py
- llamastash.json.template
- packager.py
- packager_test.py
- 2026-05-27-9806df2033dc.json
- 2026-05-27-0a82569965c3.json
- README.md
- 2026-05-25-98506aedc023.json
- 2026-05-28-9806df2033dc.json
- 2026-05-27-0a82569965c3.json
- results.md
- small-agent_decode-decode.svg
- small-agent_decode-ttft.svg
- small-chat_turn-decode.svg
- small-chat_turn-ttft.svg
- large_dense-agent_decode-decode.svg
- large_dense-agent_decode-ttft.svg
- large_dense-chat_turn-decode.svg
- large_dense-chat_turn-ttft.svg
- large_dense-parallel_4-decode.svg
- large_dense-parallel_4-ttft.svg
- large_dense-rag_prefill-decode.svg
- large_dense-rag_prefill-ttft.svg
- large_moe-agent_decode-decode.svg
- large_moe-agent_decode-ttft.svg
- large_moe-chat_turn-decode.svg
- large_moe-chat_turn-ttft.svg
- large_moe-parallel_4-decode.svg
- large_moe-parallel_4-ttft.svg
- large_moe-rag_prefill-decode.svg
- large_moe-rag_prefill-ttft.svg
- mid-agent_decode-decode.svg
- mid-agent_decode-ttft.svg
- mid-chat_turn-decode.svg
- mid-chat_turn-ttft.svg
- mid-parallel_4-decode.svg
- mid-parallel_4-ttft.svg
- mid-rag_prefill-decode.svg
- mid-rag_prefill-ttft.svg
- small-agent_decode-decode.svg
- small-agent_decode-ttft.svg
- small-chat_turn-decode.svg
- small-chat_turn-ttft.svg
- small-parallel_4-decode.svg
- small-parallel_4-ttft.svg
- small-rag_prefill-decode.svg
- small-rag_prefill-ttft.svg
- 2026-05-23-335deaf47ccf.json
- 2026-05-24-135416-0f23808f9b5f.json
- 2026-05-24-150131-0f23808f9b5f.json
- 2026-05-24-152728-0f23808f9b5f.json
- 2026-05-24-large_dense-mixed-power.json
- 2026-05-24-191126-0f23808f9b5f.json
- 2026-05-24-87c8260fa034.json
- 2026-05-24-87c8260fa034.json
- 2026-05-24-224255-deafb80af02a.json
- 2026-05-24-212347-e5814c2e7b66.json
- 2026-05-24-40c9ac84d987.json
- 2026-05-24-40c9ac84d987.json
- 2026-06-01-122455-f71bb3e445dc.json
- 2026-06-01-134142-f71bb3e445dc.json
- 2026-06-01-134142-f71bb3e445dc.json
- 2026-05-27-213320-9806df2033dc.json
- 2026-05-28-054627-9806df2033dc.json
- 2026-05-28-055402-9806df2033dc.json
- 2026-05-28-072734-9806df2033dc.json
- 2026-05-28-073332-9806df2033dc.json
- 2026-05-28-073616-9806df2033dc.json
- 2026-05-28-073932-9806df2033dc.json
- 2026-05-28-074333-9806df2033dc.json
- 2026-05-28-092711-9806df2033dc.json
- 2026-05-28-092944-9806df2033dc.json
- 2026-05-28-093243-9806df2033dc.json
- 2026-06-01-112136-591187f77dc4.json
- 2026-06-01-112212-591187f77dc4.json
- 2026-06-01-111849-591187f77dc4.json
- 2026-06-01-112010-591187f77dc4.json
- 2026-06-01-112323-591187f77dc4.json
- 2026-06-01-113730-591187f77dc4.json
- 2026-06-01-112257-591187f77dc4.json
- 2026-06-01-114714-591187f77dc4.json
- 2026-06-01-114851-591187f77dc4.json
- 2026-06-01-114837-591187f77dc4.json
- 2026-05-27-215040-0a82569965c3.json
- 2026-05-27-215807-0a82569965c3.json
- 2026-05-27-220836-0a82569965c3.json
- 2026-05-27-223618-0a82569965c3.json
- 2026-05-27-223954-0a82569965c3.json
- 2026-05-27-224142-0a82569965c3.json
- session-summary-2026-05-27.md
- README.md
- index.md
- linux-amd-apu-final-report.md
- linux-nvidia-final.md
- macos-m1-final-report.md
- methodology.md
- results-2026-05-23.md
- results-2026-05-23.preamble.md
- results-2026-05-24.md
- results-2026-05-24.preamble.md
- 2026-05-18-init-wizard-requirements.md
- 2026-05-19-release-setup-requirements.md
- 2026-05-19-uat-e2e-hardware-strategy-requirements.md
- 2026-05-20-arch-defaults-typed-advanced-editor-requirements.md
- 2026-05-20-hf-pull-tui-dialog-requirements.md
- 2026-05-21-benchmark-harness-requirements.md
- 2026-05-21-proxy-router-requirements.md
- 2026-05-29-windows-support-and-http-ipc-requirements.md
- 2026-05-31-npu-backend-via-lemonade-requirements.md
- 2026-06-06-config-presets-per-model.md
- 2026-06-08-multi-backend-abstraction-requirements.md
- 2026-06-09-lan-exposed-proxy-auth-requirements.md
- 2026-06-12-auto-fit-and-hardware-truth-requirements.md
- llamatui-requirements.md
- 2026-05-13-001-feat-llamatui-v1-launcher-plan.md
- 2026-05-16-001-feat-kdash-style-dashboard-ui-plan.md
- 2026-05-18-001-feat-init-wizard-doctor-pull-plan.md
- 2026-05-19-001-feat-interactive-init-wizard-and-cli-colors-plan.md
- 2026-05-19-001-feat-vendor-benchmark-scrapers-plan.md
- 2026-05-19-002-feat-uat-e2e-hardware-strategy-plan.md
- 2026-05-19-003-feat-0.2.0-release-setup-plan.md
- 2026-05-20-001-feat-live-hf-snapshot-discovery-plan.md
- 2026-05-20-002-feat-colorful-cli-output-plan.md
- 2026-05-20-002-feat-hf-pull-tui-dialog-plan.md
- 2026-05-20-003-feat-arch-defaults-typed-editor-plan.md
- 2026-05-21-001-feat-benchmark-harness-plan.md
- 2026-05-21-001-feat-proxy-router-plan.md
- 2026-05-29-001-feat-windows-support-and-http-ipc-plan.md
- 2026-06-05-001-feat-device-selection-plan.md
- 2026-06-08-001-refactor-backend-trait-abstraction-plan.md
- 2026-06-09-001-feat-lan-exposed-proxy-auth-plan.md
- 2026-06-10-001-feat-lemonade-backend-reintegration-plan.md
- 2026-06-13-001-feat-auto-fit-launch-mode-and-hardware-truth-plan.md
- 2026-06-15-001-feat-proxy-ui-surface-plan.md
- 2026-06-20-001-refactor-codebase-audit-improvements-plan.md
- 2026-06-22-001-feat-config-presets-per-model-plan.md
- 2026-06-24-001-feat-shared-safetensors-discovery-substrate-plan.md
- 2026-06-24-002-feat-mlx-backend-plan.md
- 2026-06-30-001-feat-default-preset-resolver-layer-plan.md
- 2026-07-10-001-feat-ds4-backend-plan.md
- 2026-07-14-001-feat-mtp-speculative-decoding-plan.md
- 2026-07-15-001-refactor-backend-config-grouping-plan.md
- 2026-07-16-001-feat-server-abstraction-plan.md
- 2026-08-09-001-feat-cli-catalog-surface-parity-plan.md
- 2026-08-10-001-feat-vllm-backend-plan.md
- pr-63.md
- review-2026-05-18-project-audit.md
- review-2026-05-24.md
- review.md
- measure-vram-overhead-band.md
- proxy-latency-bench.md
- release-0.0.1-bootstrap.md
- release-0.0.1-checklist.md
- verify-uat-reintroduction.md
- 2026-05-19-brew-linux-bottle.md
- 2026-05-19-hf-hub-client-injection.md
- 2026-05-19-hf-hub-resume.md
- 2026-05-19-llama-cpp-releases-asset-contract.md
- 2026-05-19-vram-overhead-band.md
- 2026-05-17-render-issues.md
- 2026-05-17-render-test-plan.md
- 2026-05-30-e2e-uat-plan.md
- 2026-06-01-e2e-uat-run-macos.md
- 2026-06-01-e2e-uat-windows-run.md
- 2026-06-17-e2e-uat-windows-run.md
- 2026-07-19-e2e-uat-findings.md
- 2026-07-19-e2e-uat-plan.md
- 2026-08-14-e2e-pr63-vllm-plan.md
- coverage.md
- hardware-uat.md
- architecture.md
- benchmarks.md
- lemonade-setup.md
- troubleshooting.md
- usage.md
- vllm-setup.md
- rag_prefill_8k.txt
- __init__.py
- base.py
- llamacpp.py
- llamastash.py
- lmstudio.py
- ollama.py
- __init__.py
- test_drivers_smoke.py
- test_gpu_sampler_compat.py
- test_metrics.py
- test_ollama_import_sha.py
- test_orchestrator_loop.py
- test_provenance.py
- test_render.py
- test_schema.py
- test_table.py
- test_workloads.py
- __init__.py
- charts.py
- gpu_sampler.py
- metrics.py
- model_resolver.py
- orchestrator.py
- provenance.py
- render.py
- run.sh
- schema.py
- table.py
- workloads.py
- __init__.py
- test_argv_equivalence.py
- test_threshold_classifier.py
- __init__.py
- orchestrator.py
- run.sh
- thresholds.json
- __init__.py
- orchestrator.py
- run.sh
- __init__.py
- conftest.py
- ds4_dspark_charged.sh
- dspark_ab.sh
- mtp_ab.sh
- README.md
- __init__.py
- hf_discovery.py
- hf_discovery_test.py
- README.md
- whichllm.py
- whichllm_combined.py
- cast_frames.py
- demo.prog
- drive_init_search.py
- example.prog
- harness.py
- presets.prog
- README.md
- seed_demo_models.py
- trim_cast.py
- tui_drive.py
- probe.sh
- README.md
- uat.sh
- audit_summary.py
- install.ps1
- install.sh
- install.test.bats
- measure-overhead-band.py
- measure-overhead-band.sh
- regenerate-benchmark-snapshot.py
- requirements.txt
- run-uat-and-bench.md
- run-uat-and-bench.sh
- commands.md
- SKILL.md
- mod.rs
- backend.rs
- client.rs
- discovery.rs
- mod.rs
- orchestrate.rs
- actuals.rs
- compose.rs
- list_devices.rs
- mod.rs
- telemetry.rs
- discovery.rs
- mod.rs
- identity.rs
- mod.rs
- server.rs
- isolation.rs
- lifecycle.rs
- mod.rs
- model.rs
- report.rs
- cli_args.rs
- client.rs
- colors.rs
- config.rs
- daemon.rs
- doctor.rs
- exit_codes.rs
- favorites.rs
- format.rs
- init.rs
- knob_flags.rs
- last_params.rs
- list.rs
- logs.rs
- mod.rs
- output.rs
- picker.rs
- presets.rs
- pull.rs
- resolve.rs
- show.rs
- start.rs
- status.rs
- stop.rs
- tail_args.rs
- loader.rs
- mod.rs
- presets_writer.rs
- writer.rs
- yaml_edit.rs
- actuals.rs
- auth.rs
- context.rs
- control_plane.rs
- discovery_task.rs
- host_metrics.rs
- idle.rs
- launch_service.rs
- lockfile.rs
- mod.rs
- orphans.rs
- ports.rs
- preset_store.rs
- probe.rs
- registry.rs
- resources.rs
- runtime_file.rs
- shutdown.rs
- state_store.rs
- supervisor.rs
- catalog.rs
- hf_repos.rs
- known_caches.rs
- lm_studio.rs
- metadata_cache.rs
- mod.rs
- ollama.rs
- scanner.rs
- shard_sizes.rs
- split_gguf.rs
- watcher.rs
- errors.rs
- header.rs
- identity.rs
- memory.rs
- metadata.rs
- mod.rs
- test_fixtures.rs
- amd.rs
- dxgi.rs
- metal.rs
- mod.rs
- nvidia.rs
- sysfs.rs
- vulkan.rs
- aider.rs
- claude_code.rs
- continue_dev.rs
- env_sh.rs
- mod.rs
- opencode.rs
- pi_dev.rs
- zed.rs
- merge.rs
- mod.rs
- write.rs
- brew.rs
- custom_path.rs
- gh_releases.rs
- mod.rs
- safe_extract.rs
- benchmark.rs
- config_writer.rs
- detection.rs
- doctor.rs
- download.rs
- fetch.rs
- fetch_policy.rs
- hf_api.rs
- mod.rs
- prompts.rs
- recommender.rs
- smoke.rs
- snapshot.rs
- wizard.rs
- client.rs
- methods.rs
- mod.rs
- protocol.rs
- status.rs
- admission.rs
- AGENTS.md
- binary.rs
- CLAUDE.md
- defaults_table.rs
- favorites.rs
- flag_aliases.rs
- headroom.rs
- mod.rs
- mode.rs
- native_knobs.rs
- params.rs
- presets.rs
- resolve.rs
- auth.rs
- coalesce.rs
- eviction.rs
- failure_tracker.rs
- forward.rs
- launch.rs
- mod.rs
- mru.rs
- ollama_compat.rs
- openai.rs
- route.rs
- router.rs
- server.rs
- state.rs
- ui.rs
- custom.rs
- gruvbox.rs
- latte.rs
- macchiato.rs
- mod.rs
- mono.rs
- palette.rs
- solarized.rs
- chat.rs
- embed.rs
- input_pane.rs
- logs.rs
- mod.rs
- rerank.rs
- settings.rs
- app.rs
- confirm_overlay.rs
- delete.rs
- download_strip.rs
- events.rs
- filter.rs
- fmt.rs
- glyphs.rs
- help_bar.rs
- help_overlay.rs
- hf_dialog.rs
- hf_pull.rs
- hint_picker.rs
- host_stats_pane.rs
- info_pane.rs
- input_field.rs
- keybindings.rs
- launch_picker.rs
- layout.rs
- list_pane.rs
- logo_pane.rs
- mod.rs
- oai_client.rs
- render.rs
- right_pane.rs
- save_preset_dialog.rs
- status_icons.rs
- atomic_write.rs
- clipboard.rs
- config_patch.rs
- datetime.rs
- file_security.rs
- hex.rs
- http_auth.rs
- logging.rs
- mod.rs
- model_caches.rs
- paths.rs
- process.rs
- process_control.rs
- test_temp.rs
- banner.rs
- lib.rs
- main.rs
- test_support.rs
- fake_ds4_server.rs
- fake_lemond.rs
- fake_llama_server.rs
- fake_vllm_server.rs
- chat-view.txt
- dashboard-overview.txt
- hf-confirm.txt
- hf-files.txt
- hf-search.txt
- logs-view.txt
- cli_config_test.rs
- cli_doctor_basic.rs
- cli_init_parse.rs
- cli_integration_test.rs
- control_plane_handshake_test.rs
- daemon_config_integration_test.rs
- daemon_lifecycle_test.rs
- data_yamls_well_formed.rs
- discovery_scan_test.rs
- ds4_backend_test.rs
- fetch_redirect_policy.rs
- gguf_parse_test.rs
- init_orchestration.rs
- init_snapshot_persistence.rs
- known_caches_test.rs
- lemonade_route_test.rs
- lemonade_umbrella_test.rs
- list_devices_fixture.rs
- list_models_test.rs
- preset_config_ipc_test.rs
- proxy_autostart.rs
- proxy_coalesce.rs
- proxy_echo_verification.rs
- proxy_eviction.rs
- proxy_fallback.rs
- proxy_listener_test.rs
- proxy_manual_test_plan.md
- proxy_models.rs
- proxy_ollama_compat.rs
- proxy_real_client_smoke.md
- proxy_routing.rs
- proxy_ui.rs
- recommender_corpus.rs
- split_gguf_test.rs
- start_model_ipc_test.rs
- supervisor_ipc_test.rs
- supervisor_lifecycle_test.rs
- tui_ascii_render_test.rs
- tui_chat_smoke_test.rs
- tui_e2e_render_test.rs
- tui_smoke_test.rs
- uat_lifecycle_test.rs
- vllm_backend_test.rs
- .gitignore
- AGENTS.md
- Cargo.lock
- Cargo.toml
- CHANGELOG.md
- CLAUDE.md
- config.example.yaml
- CONTRIBUTING.md
- FEATURES.md
- INSTALL.md
- LICENSE
- Makefile
- NOTICE
- README.md
- rustfmt.toml
- SECURITY.md
- TODO.md
# Installation Guide
git clone https://github.com/llamastash/llamastash
Downloads the entire project code from GitHub to your computer.
cd llamastash
Moves into the project folder you just downloaded.
2. Official Install Script
Easy Recommended- Homebrew A package manager for macOS/Linux.
- Rust (rustup) Rust is required to use cargo.
curl -fsSL https://llamastash.dev/install.sh | sh
Downloads and runs the official install script in one line β this handles the full setup automatically.
irm https://llamastash.dev/install.ps1 | iex
Downloads and runs the official install script via PowerShell β this handles the full setup automatically.
brew install llamastash/llamastash/llamastash
Installs the pre-built package via Homebrew β no source build required.
cargo install llamastash
Builds and installs the package published on crates.io β no need to clone the repo yourself.
Pulled directly from this repo's README.
3. Python
Easypip install -r scripts/requirements.txt
Installs the Python libraries listed in requirements.txt (or similar).
python <μ€νν νμΌλͺ
>.py # READMEμμ μ νν μ€ν νμΌλͺ
μ νμΈνμΈμ
Runs the Python script (or module).
4. Rust
Medium- Git Needed to download the project code from GitHub.
- Rust (rustup) Installing via rustup also installs cargo.
cargo install llamastash
Builds and installs the package published on crates.io β no need to clone the repo yourself.
Pulled directly from this repo's README.
5. Make
Medium- Git Needed to download the project code from GitHub.
- Make Usually pre-installed on Linux/macOS. On Windows, install separately (e.g. via MSYS2 or WSL).
make
Compiles the code based on the generated build configuration to produce an executable.
