summaryrefslogtreecommitdiff
path: root/Dockerfile
Commit message (Collapse)AuthorAgeFilesLines
* build(server): split Dockerfiles, make ML model paths optional (#120)soryu2026-05-021-32/+51
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Existing Dockerfile (with LLM/STT/TTS model download) is now `Dockerfile.full`. The new top-level `Dockerfile` builds a slim image without python, without huggingface_hub, without the model download step. The slim image is the new default for users who only want the orchestration surface — the directive folder UI, the mesh/task system, the API. ## Slim Dockerfile * No python / huggingface_hub / model downloads. * Same runtime tooling as `k8s/daemon/Dockerfile` (git, gh CLI, ssh, jq, curl, ca-certs, libssl3). * Embeds the daemon binary at /app/daemon-binaries/makima-linux-x86_64 for the in-server download endpoint. * PARAKEET_MODEL_DIR / SORTFORMER_MODEL_PATH / CHATTERBOX_MODEL_DIR are intentionally NOT set — Listen and Speak return "ML models not configured" if a client tries to use them. ## ML model paths now optional `ServerArgs.parakeet_model_dir`, `parakeet_eou_dir`, `sortformer_model_path`, `chatterbox_model_dir` are now `Option<String>` (no defaults). The bin constructor inspects them: if all four are present, configures `AppState::new`; if all four are absent, uses the new `AppState::new_slim()` which leaves `model_config = None`. The lazy load path in `get_ml_models` already returned a clean error for None. Speak (TTS) was already optional via `model_config.as_ref()` — still works. Mixed configurations log a warning and degrade to slim mode. ## Ops note The old `Dockerfile.full` retains the original behaviour for anyone who needs STT/diarization/TTS in production. CI still builds the daemon image from `k8s/daemon/Dockerfile` (untouched). Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* feat: Add daemon health monitoring page, downloads & K8s support (#76)soryu2026-02-211-0/+5
| | | | | | | | | | | * feat: soryu-co/soryu - makima: Add server-side daemon binary download endpoint * feat: soryu-co/soryu - makima: Create Kubernetes daemon manifests and Dockerfile * feat: soryu-co/soryu - makima: Create dedicated Daemons page with health monitoring UI * WIP: heartbeat checkpoint * feat: soryu-co/soryu - makima: Integrate daemon platform availability into frontend downloads
* Use chatterbox TTSsoryu2026-02-011-2/+2
|
* Download vocab.json and merges.txt in container imagesoryu2026-01-301-5/+0
|
* Support both tokenizor.json and vocab.json+merges.txt formatssoryu2026-01-301-0/+6
|
* Fix model loading for TTS / speaksoryu2026-01-291-1/+1
|
* fix: Add Qwen3-TTS model download to Docker build (#44)soryu2026-01-291-0/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | * chore: fix unused import warnings in qwen3-tts module - Remove unused import 'IndexOp' in model.rs - Remove unused import 'DType' in speech_tokenizer.rs - Add #[allow(dead_code)] to codebook_dim field in RvqCodebook Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * feat: add voice loading and selection for TTS cloning Add voice reference audio loading so the TTS speak handler can perform voice cloning using reference WAV files from the voices/ directory. - Add voice.rs module: loads manifest.json and reference.wav for a given voice_id, decodes via symphonia, resamples to 24kHz for the TTS engine - Update speak.rs: resolve voice_id from the speak request (default "makima"), load reference audio, pass it to engine.generate() - Add voices/makima/README.md with instructions for obtaining reference audio (extraction from YouTube, recording, ffmpeg conversion) - Graceful fallback: if reference audio is missing, TTS proceeds without voice cloning using the model's default voice Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * feat: add inference cancellation support for TTS generation Add cooperative cancellation via Arc<AtomicBool> cancel flag that threads through TtsEngine::generate -> Qwen3Tts -> GenerationContext. The autoregressive loop and streaming decoder check the flag each iteration and break early when set. The speak WebSocket handler creates a per-session flag, passes it to generate, and sets it on Cancel/Stop/Close messages. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * Add Qwen3-TTS model download to build process Fix TTS engine failure due to missing tokenizer by downloading Qwen3-TTS models during Docker build: - Download model.safetensors, config.json, tokenizer.json, and tokenizer_config.json from Qwen/Qwen3-TTS-12Hz-0.6B-Base - Download speech tokenizer from Qwen/Qwen3-TTS-Tokenizer-12Hz - Add QWEN3_TTS_DIR environment variable to Dockerfile - Script supports both env var override and default path Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
* Fixup: Add cleanup and isolation features to makimasoryu2026-01-151-3/+3
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Add comprehensive CLI documentation - Create makima/docs/CLI.md with complete command reference for: - makima server: HTTP/WebSocket server options - makima daemon: Worker daemon configuration - makima supervisor: Contract orchestration commands - makima contract: Task-contract interaction commands - Include configuration file examples and environment variables - Add usage workflows for common scenarios - Update makima/README.md with CLI overview and link to docs Add GitHub Actions release workflow for v0.1.0 Creates automated release workflow that: - Triggers on v* tag pushes - Builds binaries for Linux x86_64, macOS x86_64, and macOS ARM64 - Uses Rust nightly toolchain (required for edition 2024) - Packages binaries as .tar.gz archives - Creates GitHub release with installation instructions fix(ci): update macOS runner for x86_64 builds Replace deprecated macos-13 runner with macos-15-intel for x86_64-apple-darwin target. The macos-13 runner has been retired by GitHub Actions. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> Add dismissing notifications and fix CLI task ID arg Add worktree cleanup when contracts complete or are deleted (#21) - Add CleanupWorktree daemon command variant - Handle CleanupWorktree in daemon task manager - Add cleanup_contract_worktrees helper function - Trigger cleanup when contract status becomes 'completed' - Trigger cleanup before contract deletion Add Autonomous Loop Mode for persistent task completion (#20) Implements the "Autonomous Loop Mode" feature inspired by Ralph for Claude Code. This enables tasks to automatically restart and continue working until they explicitly signal completion via a COMPLETION_GATE block. Key features: - Exit confirmation via COMPLETION_GATE: Tasks must output a <COMPLETION_GATE> block with `ready: true` to signal completion. Without this, the task auto-restarts using `claude --continue` to resume the conversation. - Circuit breaker: Prevents infinite loops by detecting: * Maximum iteration limit (default: 10) * No progress for N consecutive iterations (default: 3) * Same error repeated N times (default: 5) - spawn_continue: New ProcessManager method to spawn Claude with the `--continue` flag, resuming from the previous session state. Toggle: Enable via `autonomous_loop` flag on contracts. When set, all tasks spawned for that contract will run in autonomous loop mode. Files changed: - completion_gate.rs: COMPLETION_GATE parser and CircuitBreaker logic - claude.rs: spawn_continue() for --continue mode spawning - manager.rs: Autonomous loop iteration logic in run_task() - protocol.rs: autonomousLoop field in DaemonCommand::SpawnTask - models.rs/repository.rs: autonomous_loop column on contracts/tasks - Migration: Adds autonomous_loop columns to contracts and tasks tables Add get-task and output commands to supervisor CLI (#24) Add two new supervisor subcommands: - `makima supervisor task <task_id>` - Get individual task details - `makima supervisor output <task_id>` - Get task output/claude log This allows supervisors to fetch task details and claude output directly from the CLI instead of using curl to call the task API. Add optional bubblewrap sandboxing for Claude processes (#23) Add --bubblewrap flag and process.bubblewrap config section to enable running Claude Code in a bubblewrap sandbox for process isolation. When enabled, claude processes run with filesystem restrictions: - Root filesystem mounted read-only - Working directory (worktree) mounted read-write - Fresh /dev, /proc, /tmp - Network access preserved for API calls Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
* Fixup: experiment with different entrypoint for containersoryu2026-01-151-1/+1
|
* Fixup: container image binarysoryu2026-01-151-4/+2
|
* Add cleanup to daemonsoryu2026-01-151-1/+2
| | | | Also fixup for container image
* Fixup: use entrypoint in containersoryu2026-01-151-5/+6
|
* Fixup: make makima CLI executablesoryu2026-01-151-2/+2
|
* Update container image to use new CLIsoryu2026-01-151-3/+3
|
* Bump diarization version to 2.1 and fix downloading the tokenizersoryu2025-12-231-1/+1
|
* Use HF to download modelssoryu2025-12-231-16/+18
|
* Fixup: Use single stage dockerfilesoryu2025-12-231-14/+2
|
* Fixup: Add stt-client modulesoryu2025-12-231-0/+1
|
* Fixup: Bump Rust & Cargo version to match versionsoryu2025-12-231-1/+1
|
* Fixup: Move container image to build from rootsoryu2025-12-231-0/+58
generated by cgit v1.2.3 (git 2.39.1) at 2026-05-20 22:28:02 +0000