agent-cli 0.61.5__tar.gz → 0.63.0__tar.gz

{agent_cli-0.61.5 → agent_cli-0.63.0}/.pre-commit-config.yaml

@@ -8,7 +8,7 @@ repos:
       - id: end-of-file-fixer
       - id: mixed-line-ending
   - repo: https://github.com/astral-sh/ruff-pre-commit
-    rev: "v0.14.11"
+    rev: "v0.14.13"
     hooks:
       - id: ruff
         args: ["--fix"]

{agent_cli-0.61.5 → agent_cli-0.63.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: agent-cli
-Version: 0.61.5
+Version: 0.63.0
 Summary: A suite of AI-powered command-line tools for text correction, audio transcription, and voice assistance.
 Project-URL: Homepage, https://github.com/basnijholt/agent-cli
 Author-email: Bas Nijholt <bas@nijho.lt>
@@ -128,7 +128,7 @@ Since then I have expanded the tool with many more features, all focused on loca
 - **[`memory`](docs/commands/memory.md)**: Long-term memory system with `memory proxy` and `memory add`.
 - **[`rag-proxy`](docs/commands/rag-proxy.md)**: RAG proxy server for chatting with your documents.
 - **[`dev`](docs/commands/dev.md)**: Parallel development with git worktrees and AI coding agents.
-- **[`server`](docs/commands/server.md)**: Local ASR and TTS servers with dual-protocol (Wyoming & OpenAI), TTL-based memory management, and multi-platform acceleration. Whisper uses MLX on Apple Silicon or Faster Whisper on Linux/CUDA. TTS supports Kokoro (GPU) or Piper (CPU).
+- **[`server`](docs/commands/server/index.md)**: Local ASR and TTS servers with dual-protocol (Wyoming & OpenAI), TTL-based memory management, and multi-platform acceleration. Whisper uses MLX on Apple Silicon or Faster Whisper on Linux/CUDA. TTS supports Kokoro (GPU) or Piper (CPU).
 - **[`transcribe-daemon`](docs/commands/transcribe-daemon.md)**: Continuous background transcription with VAD. Install with `uv tool install "agent-cli[vad]"`.
 
 ## Quick Start
@@ -389,7 +389,7 @@ Our installation scripts automatically handle all dependencies:
 |---------|---------|-----------------|
 | **[Ollama](https://ollama.ai/)** | Local LLM for text processing | ✅ Yes, with default model |
 | **[Wyoming Faster Whisper](https://github.com/rhasspy/wyoming-faster-whisper)** | Speech-to-text | ✅ Yes, via `uvx` |
-| **[`agent-cli server whisper`](docs/commands/server.md#whisper)** | Speech-to-text (alternative) | ✅ Built-in, `pip install "agent-cli[whisper]"` |
+| **[`agent-cli server whisper`](docs/commands/server/whisper.md)** | Speech-to-text (alternative) | ✅ Built-in, `pip install "agent-cli[whisper]"` |
 | **[Wyoming Piper](https://github.com/rhasspy/wyoming-piper)** | Text-to-speech | ✅ Yes, via `uvx` |
 | **[Kokoro-FastAPI](https://github.com/remsky/Kokoro-FastAPI)** | Premium TTS (optional) | ⚙️ Can be added later |
 | **[Wyoming openWakeWord](https://github.com/rhasspy/wyoming-openwakeword)** | Wake word detection | ✅ Yes, for `assistant` |

{agent_cli-0.61.5 → agent_cli-0.63.0}/README.md

@@ -48,7 +48,7 @@ Since then I have expanded the tool with many more features, all focused on loca
 - **[`memory`](docs/commands/memory.md)**: Long-term memory system with `memory proxy` and `memory add`.
 - **[`rag-proxy`](docs/commands/rag-proxy.md)**: RAG proxy server for chatting with your documents.
 - **[`dev`](docs/commands/dev.md)**: Parallel development with git worktrees and AI coding agents.
-- **[`server`](docs/commands/server.md)**: Local ASR and TTS servers with dual-protocol (Wyoming & OpenAI), TTL-based memory management, and multi-platform acceleration. Whisper uses MLX on Apple Silicon or Faster Whisper on Linux/CUDA. TTS supports Kokoro (GPU) or Piper (CPU).
+- **[`server`](docs/commands/server/index.md)**: Local ASR and TTS servers with dual-protocol (Wyoming & OpenAI), TTL-based memory management, and multi-platform acceleration. Whisper uses MLX on Apple Silicon or Faster Whisper on Linux/CUDA. TTS supports Kokoro (GPU) or Piper (CPU).
 - **[`transcribe-daemon`](docs/commands/transcribe-daemon.md)**: Continuous background transcription with VAD. Install with `uv tool install "agent-cli[vad]"`.
 
 ## Quick Start
@@ -309,7 +309,7 @@ Our installation scripts automatically handle all dependencies:
 |---------|---------|-----------------|
 | **[Ollama](https://ollama.ai/)** | Local LLM for text processing | ✅ Yes, with default model |
 | **[Wyoming Faster Whisper](https://github.com/rhasspy/wyoming-faster-whisper)** | Speech-to-text | ✅ Yes, via `uvx` |
-| **[`agent-cli server whisper`](docs/commands/server.md#whisper)** | Speech-to-text (alternative) | ✅ Built-in, `pip install "agent-cli[whisper]"` |
+| **[`agent-cli server whisper`](docs/commands/server/whisper.md)** | Speech-to-text (alternative) | ✅ Built-in, `pip install "agent-cli[whisper]"` |
 | **[Wyoming Piper](https://github.com/rhasspy/wyoming-piper)** | Text-to-speech | ✅ Yes, via `uvx` |
 | **[Kokoro-FastAPI](https://github.com/remsky/Kokoro-FastAPI)** | Premium TTS (optional) | ⚙️ Can be added later |
 | **[Wyoming openWakeWord](https://github.com/rhasspy/wyoming-openwakeword)** | Wake word detection | ✅ Yes, for `assistant` |

{agent_cli-0.61.5 → agent_cli-0.63.0}/agent_cli/agents/assistant.py

@@ -253,7 +253,7 @@ async def _async_main(
                 print_with_style("✨ Ready for next command...", style="green")
 
 
-@app.command("assistant")
+@app.command("assistant", rich_help_panel="Voice Commands")
 def assistant(
     *,
     # --- Provider Selection ---

{agent_cli-0.61.5 → agent_cli-0.63.0}/agent_cli/agents/autocorrect.py

@@ -208,7 +208,7 @@ async def _async_autocorrect(
         sys.exit(1)
 
 
-@app.command("autocorrect")
+@app.command("autocorrect", rich_help_panel="Text Commands")
 def autocorrect(
     *,
     text: str | None = typer.Argument(

{agent_cli-0.61.5 → agent_cli-0.63.0}/agent_cli/agents/chat.py

@@ -373,7 +373,7 @@ async def _async_main(
         raise
 
 
-@app.command("chat")
+@app.command("chat", rich_help_panel="Voice Commands")
 def chat(
     *,
     # --- Provider Selection ---

{agent_cli-0.61.5 → agent_cli-0.63.0}/agent_cli/agents/memory/__init__.py

@@ -14,7 +14,7 @@ memory_app = typer.Typer(
     no_args_is_help=True,
 )
 
-app.add_typer(memory_app, name="memory")
+app.add_typer(memory_app, name="memory", rich_help_panel="Servers")
 
 
 @memory_app.callback()

{agent_cli-0.61.5 → agent_cli-0.63.0}/agent_cli/agents/rag_proxy.py

@@ -17,7 +17,7 @@ from agent_cli.core.utils import (
 )
 
 
-@app.command("rag-proxy")
+@app.command("rag-proxy", rich_help_panel="Servers")
 def rag_proxy(
     docs_folder: Path = typer.Option(  # noqa: B008
         "./rag_docs",

{agent_cli-0.61.5 → agent_cli-0.63.0}/agent_cli/agents/speak.py

@@ -79,7 +79,7 @@ async def _async_main(
     return text
 
 
-@app.command("speak")
+@app.command("speak", rich_help_panel="Text Commands")
 def speak(
     *,
     text: str | None = typer.Argument(

{agent_cli-0.61.5 → agent_cli-0.63.0}/agent_cli/agents/transcribe.py

@@ -460,7 +460,7 @@ async def _async_main(  # noqa: PLR0912, PLR0915, C901
     )
 
 
-@app.command("transcribe")
+@app.command("transcribe", rich_help_panel="Voice Commands")
 def transcribe(  # noqa: PLR0912
     *,
     extra_instructions: str | None = typer.Option(

{agent_cli-0.61.5 → agent_cli-0.63.0}/agent_cli/agents/transcribe_daemon.py

@@ -286,7 +286,7 @@ async def _daemon_loop(cfg: DaemonConfig) -> None:  # noqa: PLR0912, PLR0915
                     await asyncio.wait(background_tasks, timeout=2.0)
 
 
-@app.command("transcribe-daemon")
+@app.command("transcribe-daemon", rich_help_panel="Voice Commands")
 def transcribe_daemon(  # noqa: PLR0912
     *,
     # Daemon-specific options

{agent_cli-0.61.5 → agent_cli-0.63.0}/agent_cli/agents/voice_edit.py

@@ -172,7 +172,7 @@ async def _async_main(
         )
 
 
-@app.command("voice-edit")
+@app.command("voice-edit", rich_help_panel="Voice Commands")
 def voice_edit(
     *,
     # --- Provider Selection ---

{agent_cli-0.61.5 → agent_cli-0.63.0}/agent_cli/dev/cli.py

@@ -122,7 +122,7 @@ app = typer.Typer(
     rich_markup_mode="markdown",
     no_args_is_help=True,
 )
-main_app.add_typer(app, name="dev")
+main_app.add_typer(app, name="dev", rich_help_panel="Development")
 
 
 @app.callback()
@@ -344,6 +344,19 @@ def _launch_editor(path: Path, editor: Editor) -> None:
         _warn(f"Could not open editor: {e}")
 
 
+def _write_prompt_to_worktree(worktree_path: Path, prompt: str) -> Path:
+    """Write the prompt to .claude/TASK.md in the worktree.
+
+    This makes the task description available to the spawned agent
+    and provides a record of what was requested.
+    """
+    claude_dir = worktree_path / ".claude"
+    claude_dir.mkdir(parents=True, exist_ok=True)
+    task_file = claude_dir / "TASK.md"
+    task_file.write_text(prompt + "\n")
+    return task_file
+
+
 def _format_env_prefix(env: dict[str, str]) -> str:
     """Format environment variables as shell prefix.
 
@@ -627,6 +640,11 @@ def new(  # noqa: PLR0912, PLR0915
             # Only warn if user explicitly requested direnv
             _warn("direnv not installed, skipping .envrc setup")
 
+    # Write prompt to worktree (makes task available to the spawned agent)
+    if prompt:
+        task_file = _write_prompt_to_worktree(result.path, prompt)
+        _success(f"Wrote task to {task_file.relative_to(result.path)}")
+
     # Resolve editor and agent
     resolved_editor = _resolve_editor(editor, editor_name, default_editor)
     resolved_agent = _resolve_agent(agent, agent_name, default_agent)
@@ -1039,6 +1057,11 @@ def start_agent(
     if not agent.is_available():
         _error(f"{agent.name} is not installed. Install from: {agent.install_url}")
 
+    # Write prompt to worktree (makes task available to the agent)
+    if prompt:
+        task_file = _write_prompt_to_worktree(wt.path, prompt)
+        _success(f"Wrote task to {task_file.relative_to(wt.path)}")
+
     merged_args = _merge_agent_args(agent, agent_args)
     agent_env = _get_agent_env(agent)
     _info(f"Starting {agent.name} in {wt.path}...")

{agent_cli-0.61.5 → agent_cli-0.63.0}/agent_cli/dev/skill/SKILL.md

@@ -47,8 +47,9 @@ agent-cli dev new <branch-name> --agent --prompt-file path/to/prompt.md
 This creates:
 1. A new git worktree with its own branch
 2. Runs project setup (installs dependencies)
-3. Opens a new terminal tab with an AI coding agent
-4. Passes your prompt to the agent
+3. Saves your prompt to `.claude/TASK.md` in the worktree (for reference)
+4. Opens a new terminal tab with an AI coding agent
+5. Passes your prompt to the agent
 
 **Important**: Use `--prompt-file` for prompts longer than a single line. The `--prompt` option passes text through the shell, which can cause issues with special characters (exclamation marks, dollar signs, backticks, quotes) in ZSH and other shells. Using `--prompt-file` avoids all shell quoting issues.
 

{agent_cli-0.61.5 → agent_cli-0.63.0}/agent_cli/dev/worktree.py

@@ -55,6 +55,15 @@ def is_git_repo(path: Path | None = None) -> bool:
         return False
 
 
+def has_origin_remote(path: Path | None = None) -> bool:
+    """Check if the repository has an 'origin' remote configured."""
+    try:
+        result = _run_git("remote", "get-url", "origin", cwd=path, check=False)
+        return result.returncode == 0
+    except Exception:
+        return False
+
+
 def get_repo_root(path: Path | None = None) -> Path | None:
     """Get the root directory of the git repository."""
     try:
@@ -461,6 +470,35 @@ def _init_submodules(
     )
 
 
+def _pull_lfs(
+    worktree_path: Path,
+    *,
+    on_log: Callable[[str], None] | None = None,
+    capture_output: bool = True,
+) -> None:
+    """Pull Git LFS files in a worktree if LFS is used.
+
+    Evidence: https://git-lfs.com/ - `git lfs pull` fetches LFS objects.
+    This is a no-op if LFS is not used or files are already present.
+    """
+    # Check if .gitattributes contains LFS filters
+    gitattributes = worktree_path / ".gitattributes"
+    if not gitattributes.exists():
+        return
+
+    if "filter=lfs" not in gitattributes.read_text():
+        return
+
+    # Check if git-lfs is installed
+    if not shutil.which("git-lfs"):
+        return
+
+    if on_log:
+        on_log("Pulling Git LFS files...")
+
+    _run_git("lfs", "pull", cwd=worktree_path, check=False, capture_output=capture_output)
+
+
 def _add_worktree(
     branch_name: str,
     worktree_path: Path,
@@ -586,8 +624,11 @@ def create_worktree(
     # Create base directory if needed
     base_dir.mkdir(parents=True, exist_ok=True)
 
-    # Fetch latest refs
-    if fetch:
+    # Check if origin remote exists
+    origin_exists = has_origin_remote(repo_root)
+
+    # Fetch latest refs (only if origin exists)
+    if fetch and origin_exists:
         if on_log:
             on_log("Running: git fetch origin")
         _run_git("fetch", "origin", cwd=repo_root, check=False, capture_output=capture_output)
@@ -596,10 +637,10 @@ def create_worktree(
     from_ref_explicit = from_ref is not None
 
     # Determine the reference to create from
-    # Use origin/{branch} to ensure we're using the freshly-fetched remote ref,
-    # not a potentially stale local branch
+    # Use origin/{branch} if origin exists, otherwise use local branch
     if from_ref is None:
-        from_ref = f"origin/{get_default_branch(repo_root)}"
+        default_branch = get_default_branch(repo_root)
+        from_ref = f"origin/{default_branch}" if origin_exists else default_branch
 
     # Check if branch exists remotely or locally
     remote_exists, local_exists = _check_branch_exists(branch_name, repo_root)
@@ -634,6 +675,13 @@ def create_worktree(
             capture_output=capture_output,
         )
 
+        # Pull Git LFS files if the repo uses LFS
+        _pull_lfs(
+            worktree_path,
+            on_log=on_log,
+            capture_output=capture_output,
+        )
+
         return CreateWorktreeResult(
             success=True,
             path=worktree_path,

{agent_cli-0.61.5 → agent_cli-0.63.0}/agent_cli/memory/_retrieval.py

@@ -25,6 +25,7 @@ LOGGER = logging.getLogger(__name__)
 
 _DEFAULT_MMR_LAMBDA = 0.7
 _SUMMARY_ROLE = "summary"
+_MIN_MAX_EPSILON = 1e-8  # Avoid division by zero in min-max normalization
 
 
 def gather_relevant_existing_memories(
@@ -135,7 +136,7 @@ def retrieve_memory(
     include_summary: bool = True,
     mmr_lambda: float = _DEFAULT_MMR_LAMBDA,
     recency_weight: float = 0.2,
-    score_threshold: float = 0.35,
+    score_threshold: float | None = None,
     filters: dict[str, Any] | None = None,
 ) -> tuple[MemoryRetrieval, list[str]]:
     """Execute search + rerank + recency + MMR."""
@@ -161,8 +162,15 @@ def retrieve_memory(
             seen_ids.add(rec_id)
             raw_candidates.append(rec)
 
-    def _sigmoid(x: float) -> float:
-        return 1.0 / (1.0 + math.exp(-x))
+    def _min_max_normalize(scores: list[float]) -> list[float]:
+        """Normalize scores to 0-1 range using min-max scaling."""
+        if not scores:
+            return scores
+        min_score = min(scores)
+        max_score = max(scores)
+        if max_score - min_score < _MIN_MAX_EPSILON:
+            return [0.5] * len(scores)  # All scores equal
+        return [(s - min_score) / (max_score - min_score) for s in scores]
 
     def recency_score(meta: MemoryMetadata) -> float:
         dt = datetime.fromisoformat(meta.created_at)
@@ -176,10 +184,12 @@ def retrieve_memory(
     if raw_candidates:
         pairs = [(query, mem.content) for mem in raw_candidates]
         rr_scores = predict_relevance(reranker_model, pairs)
-        for mem, rr in zip(raw_candidates, rr_scores, strict=False):
-            relevance = _sigmoid(rr)
-            # Filter out low-relevance memories to reduce noise
-            if relevance < score_threshold:
+        # Normalize raw reranker scores to 0-1 range
+        normalized_scores = _min_max_normalize(rr_scores)
+
+        for mem, relevance in zip(raw_candidates, normalized_scores, strict=False):
+            # Filter out low-relevance memories if threshold is set
+            if score_threshold is not None and relevance < score_threshold:
                 continue
 
             recency = recency_score(mem.metadata)
@@ -235,7 +245,7 @@ async def augment_chat_request(
     include_global: bool = True,
     mmr_lambda: float = _DEFAULT_MMR_LAMBDA,
     recency_weight: float = 0.2,
-    score_threshold: float = 0.35,
+    score_threshold: float | None = None,
     filters: dict[str, Any] | None = None,
 ) -> tuple[ChatRequest, MemoryRetrieval | None, str, list[str]]:
     """Retrieve memory context and augment the chat request."""

{agent_cli-0.61.5 → agent_cli-0.63.0}/agent_cli/memory/api.py

@@ -30,7 +30,7 @@ def create_app(
     max_entries: int = 500,
     mmr_lambda: float = 0.7,
     recency_weight: float = 0.2,
-    score_threshold: float = 0.35,
+    score_threshold: float | None = None,
     enable_git_versioning: bool = True,
 ) -> FastAPI:
     """Create the FastAPI app for memory-backed chat."""

{agent_cli-0.61.5 → agent_cli-0.63.0}/agent_cli/memory/client.py

@@ -49,7 +49,7 @@ class MemoryClient:
         max_entries: int = 500,
         mmr_lambda: float = 0.7,
         recency_weight: float = 0.2,
-        score_threshold: float = 0.35,
+        score_threshold: float | None = None,
         start_watcher: bool = False,
         enable_git_versioning: bool = True,
     ) -> None:

{agent_cli-0.61.5 → agent_cli-0.63.0}/agent_cli/memory/engine.py

@@ -236,7 +236,7 @@ async def process_chat_request(
     max_entries: int = _DEFAULT_MAX_ENTRIES,
     mmr_lambda: float = _DEFAULT_MMR_LAMBDA,
     recency_weight: float = 0.2,
-    score_threshold: float = 0.35,
+    score_threshold: float | None = None,
     postprocess_in_background: bool = True,
     enable_git_versioning: bool = False,
     filters: dict[str, Any] | None = None,

{agent_cli-0.61.5 → agent_cli-0.63.0}/agent_cli/server/cli.py

@@ -34,7 +34,7 @@ app = typer.Typer(
     rich_markup_mode="markdown",
     no_args_is_help=True,
 )
-main_app.add_typer(app, name="server")
+main_app.add_typer(app, name="server", rich_help_panel="Servers")
 
 
 @app.callback()

{agent_cli-0.61.5 → agent_cli-0.63.0}/docs/architecture/memory.md

@@ -207,8 +207,8 @@ Executed synchronously during `augment_chat_request`.
 *   **Model:** `Xenova/ms-marco-MiniLM-L-6-v2` (ONNX, quantized).
 *   **Input:** Pairs of `(user_query, memory_content)`.
 *   **Output:** Raw logit score.
-*   **Normalization:** `relevance = sigmoid(raw_score)`.
-*   **Thresholding:** Candidates with `relevance < score_threshold` (default 0.35) are hard-pruned.
+*   **Normalization:** Min-max scaling across the candidate batch (best match = 1.0, worst = 0.0).
+*   **Thresholding:** If `score_threshold` is set, candidates below it are pruned. Default is `None` (no filtering).
 
 ### Step 3: Recency Scoring
 Combines semantic relevance with temporal proximity.
@@ -309,7 +309,7 @@ To replicate the system behavior, the following prompt strategies are required.
 | `max_entries` | `500` | Hard cap on memories per conversation. |
 | `mmr_lambda` | `0.7` | Diversity weighting (1.0 = pure relevance). |
 | `recency_weight` | `0.2` | Score weight for temporal proximity. |
-| `score_threshold` | `0.35` | Minimum semantic relevance to consider. |
+| `score_threshold` | `None` | Minimum semantic relevance to consider (no filtering by default). |
 | `enable_summarization` | `True` | Toggle for summary generation loop. |
 | `openai_base_url` | *required* | Base URL for LLM calls (can point to OpenAI-compatible proxies). |
 | `enable_git_versioning` | `True` | Toggle to enable/disable Git versioning of the memory store. |

{agent_cli-0.61.5 → agent_cli-0.63.0}/docs/commands/dev.md

@@ -61,8 +61,8 @@ agent-cli dev new [BRANCH] [OPTIONS]
 | `--with-editor` | Specific editor (cursor, vscode, zed, etc.) |
 | `--with-agent` | Specific agent (claude, codex, gemini, aider) |
 | `--agent-args` | Extra arguments to pass to the agent |
-| `--prompt`, `-p` | Initial prompt to pass to the AI agent |
-| `--prompt-file`, `-P` | Read initial prompt from a file (avoids shell quoting issues) |
+| `--prompt`, `-p` | Initial prompt to pass to the AI agent (saved to `.claude/TASK.md` in worktree) |
+| `--prompt-file`, `-P` | Read initial prompt from a file (saved to `.claude/TASK.md` in worktree) |
 | `--direnv` | Generate .envrc file for direnv (auto-detects venv) |
 | `--setup/--no-setup` | Run automatic project setup (default: enabled) |
 | `--copy-env/--no-copy-env` | Copy .env files from main repo (default: enabled) |
@@ -223,8 +223,8 @@ agent-cli dev agent NAME [--agent/-a AGENT] [--agent-args ARGS] [--prompt/-p PRO
 |--------|-------------|
 | `--agent`, `-a` | Specific agent (claude, codex, gemini, aider) |
 | `--agent-args` | Extra arguments to pass to the agent |
-| `--prompt`, `-p` | Initial prompt to pass to the AI agent |
-| `--prompt-file`, `-P` | Read initial prompt from a file (avoids shell quoting issues) |
+| `--prompt`, `-p` | Initial prompt to pass to the AI agent (saved to `.claude/TASK.md` in worktree) |
+| `--prompt-file`, `-P` | Read initial prompt from a file (saved to `.claude/TASK.md` in worktree) |
 
 **Examples:**
 
@@ -495,6 +495,16 @@ git submodule update --init --recursive
 
 This ensures all submodule dependencies are available in the worktree.
 
+### Git LFS
+
+If the repository uses Git LFS (detected via `filter=lfs` in `.gitattributes`), LFS files are automatically pulled:
+
+```bash
+git lfs pull
+```
+
+This ensures large files tracked by LFS are available in the worktree.
+
 ### Environment Files
 
 The following files are automatically copied to new dev environments:

{agent_cli-0.61.5 → agent_cli-0.63.0}/docs/configuration.md

@@ -235,7 +235,7 @@ Now just run `agent-cli transcribe` - it automatically uses your local server.
 
 > [!TIP]
 > **OpenAI SDK users:** The server also exposes an OpenAI-compatible API on port 10301.
-> See [server whisper docs](commands/server.md#whisper) for all options.
+> See [server whisper docs](commands/server/whisper.md) for all options.
 
 ## Audio Device Configuration
 

{agent_cli-0.61.5 → agent_cli-0.63.0}/docs/getting-started.md

@@ -43,7 +43,7 @@ For a complete local setup with all AI services:
 > pip install "agent-cli[whisper-mlx]"
 > agent-cli server whisper --backend mlx
 > ```
-> See [Local Whisper Server](commands/server.md#whisper) for details.
+> See [Local Whisper Server](commands/server/whisper.md) for details.
 
 === "Using CLI Commands"
 

{agent_cli-0.61.5 → agent_cli-0.63.0}/docs/index.md

@@ -64,7 +64,7 @@ Since then I have expanded the tool with many more features, all focused on loca
 | [`rag-proxy`](commands/rag-proxy.md) | Chat with your documents via RAG |
 | [`memory`](commands/memory.md) | Long-term memory system for conversations |
 | [`dev`](commands/dev.md) | Parallel development with git worktrees and AI coding agents |
-| [`server`](commands/server.md) | Local ASR & TTS servers with dual-protocol (Wyoming & OpenAI), TTL-based memory, and multi-platform acceleration (MLX/CUDA) |
+| [`server`](commands/server/index.md) | Local ASR & TTS servers with dual-protocol (Wyoming & OpenAI), TTL-based memory, and multi-platform acceleration (MLX/CUDA) |
 
 ## Quick Start