crow-cli 0.1.30__tar.gz → 0.1.32__tar.gz

{crow_cli-0.1.30 → crow_cli-0.1.32}/PKG-INFO

@@ -1,24 +1,23 @@
-Metadata-Version: 2.4
+Metadata-Version: 2.5
 Name: crow-cli
-Version: 0.1.30
+Version: 0.1.32
 Summary: Add your description here
 Project-URL: Homepage, https://github.com/crow-cli/crow-cli
 Project-URL: Repository, https://github.com/crow-cli/crow-cli
 Project-URL: Issues, https://github.com/crow-cli/crow-cli/issues
 Project-URL: Documentation, https://github.com/crow-cli/crow-cli#readme
 Requires-Python: >=3.14
-Requires-Dist: agent-client-protocol>=0.9.0
+Requires-Dist: agent-client-protocol<0.13.0,>=0.12.0
 Requires-Dist: coolname>=2.2.0
-Requires-Dist: crow-memory>=0.1.30
 Requires-Dist: directory-tree>=1.0.0
+Requires-Dist: docker>=7.2.0
 Requires-Dist: fastmcp>=2.14.5
 Requires-Dist: httpx>=0.28.1
-Requires-Dist: ipython>=9.10.0
 Requires-Dist: jinja2>=3.1.6
-Requires-Dist: json-schema-to-pydantic>=0.4.9
 Requires-Dist: openai>=2.21.0
 Requires-Dist: pyyaml>=6.0.3
 Requires-Dist: rich>=14.3.3
+Requires-Dist: sqlalchemy>=2.0.51
 Requires-Dist: typer>=0.24.1
 Description-Content-Type: text/markdown
 
@@ -34,12 +33,12 @@ Description-Content-Type: text/markdown
 
 `crow-cli` is an [Agent Client Protocol (ACP)](https://agentclientprotocol.com/) coding agent that runs in your terminal and inside ACP-compatible editors. It reads and edits code, runs shell commands, searches the web, and remembers your work across sessions.
 
-Most agent toolkits treat persistence as an afterthought. crow-cli treats it as the point: every session lives in a dedicated memory service ([crow-memory](#crow-memory--persistence--memory-api)) built on [LanceDB](https://lancedb.github.io/lancedb/) with ColBERT and ColPali embeddings, so agents recall past conversations semantically and can delegate work to one another. Sessions get memorable coolname ids (like `taupe-squirrel-of-splendid-potency`) you can resume or read from any other agent.
+Most agent toolkits treat persistence as an afterthought. crow-cli treats it as the point: every session lives in a local sqlite database (`~/.agents/crow/crow.db`) with FTS5 full-text search, so agents recall past conversations and can delegate work to one another. Images are stored as files next to the database and hydrated only when sent to the LLM. Sessions get memorable coolname ids (like `taupe-squirrel-of-splendid-potency`) you can resume or read from any other agent.
 
 ## Requirements
 
 - Python 3.14+, managed with [uv](https://docs.astral.sh/uv/)
-- Docker, for the crow-memory and SearXNG services
+- Docker, for the SearXNG service
 - An API key for an OpenAI-compatible LLM provider (OpenRouter, OpenAI, your own endpoint, …)
 
 | Platform | Notes |
@@ -61,11 +60,11 @@ uv tool install crow-cli --python 3.14      # or run without installing: uvx cro
 Initialize your configuration and start the backing services:
 
 ```bash
-crow-cli init                          # scaffolds ~/.crow (config.yaml, .env, docker-compose)
-cd ~/.crow && docker compose up -d     # starts crow-memory + SearXNG
+crow-cli init                          # scaffolds ~/.agents/crow (config.yaml, .env, docker-compose)
+cd ~/.agents/crow && docker compose up -d     # starts SearXNG
 ```
 
-`crow-cli init` walks you through provider and model selection and writes your secrets to `~/.crow/.env`, referenced from the config as `${VAR}`.
+`crow-cli init` walks you through provider and model selection and writes your secrets to `~/.agents/crow/.env`, referenced from the config as `${VAR}`.
 
 ## Quick start
 
@@ -113,17 +112,17 @@ crow-cli is a monorepo. The pieces:
 
 ### crow-cli — the agent
 
-The ACP-native agent: a streaming ReAct loop with tool calling, cancellation, conversation compaction, and multimodal input. Provider and model configuration lives in `~/.crow/config.yaml`.
+The ACP-native agent: a streaming ReAct loop with tool calling, cancellation, conversation compaction, and multimodal input. Provider and model configuration lives in `~/.agents/crow/config.yaml`.
 
-### crow-memory — persistence + memory API
+### Persistence — sqlite memory
 
-A standalone service — a LanceDB store with ColBERT (text) and ColPali (image) multivector embeddings — that the agent talks to over HTTP. It backs both session persistence and a semantic memory API, exposed to agents as three tools:
+Sessions persist to a single sqlite database (`~/.agents/crow/crow.db`, schema v3, WAL mode) with an FTS5 index for BM25 keyword search. Images in messages are written to `~/.agents/crow/images/` and referenced by path; they are hydrated to base64 data URLs only when the conversation is sent to the LLM. The same database backs the memory API, exposed to agents as three tools:
 
 - `list_sessions()` — sessions ordered by recent activity (who's working on what)
 - `query_memory(query)` — find which session discussed something, across all sessions
 - `query_session(session_id)` — read or search within one session (spans all of that session's agents)
 
-This is what makes multi-agent delegation work: launch a worker, then read its thoughts from any other agent. Today crow-memory runs as a Docker container the agent connects to; longer-term it moves toward an always-on daemon, in line with the ACP v2 direction.
+This is what makes multi-agent delegation work: launch a worker, then read its thoughts from any other agent. No service to run — the sqlite file is the integration point.
 
 ### crow-mcp — the tool server
 
@@ -137,7 +136,7 @@ The built-in [MCP](https://modelcontextprotocol.io/) server providing the agent'
 | `capture_webcam` / `read_image_file` | Vision input |
 | `list_sessions` / `query_memory` / `query_session` | Memory (see above) |
 
-**Extensible by design:** register any MCP server in `~/.crow/config.yaml` and its tools appear alongside these automatically.
+**Extensible by design:** register any MCP server in `~/.agents/crow/config.yaml` and its tools appear alongside these automatically.
 
 > ⚠️ **Tool names are not namespaced.** crow-mcp registers its tools as `read`, `edit`, `terminal`, … — not `crow-mcp_read`. When you add your own MCP servers, watch for name collisions.
 
@@ -147,11 +146,11 @@ crow-cli ships a maintained SearXNG configuration (stored as JSON so the agent c
 
 ### Skills
 
-Agents load reusable skills from `~/.crow/skills/` — each a directory with a `SKILL.md` describing when and how to use it. Skill distribution is still being worked out; today skills are local directories.
+Agents load reusable skills from `~/.agents/skills/` — each a directory with a `SKILL.md` describing when and how to use it. Skill distribution is still being worked out; today skills are local directories.
 
 ## Configuration
 
-`~/.crow/config.yaml` holds providers, models, and MCP servers; secrets live in `~/.crow/.env` and are interpolated with `${VAR}`.
+`~/.agents/crow/config.yaml` holds providers, models, and MCP servers; secrets live in `~/.agents/crow/.env` and are interpolated with `${VAR}`.
 
 ```yaml
 providers:
@@ -178,7 +177,7 @@ Run the unit tests — fast and hermetic, no services required (tests that touch
 uv run --project crow-cli pytest crow-cli/tests/unit
 ```
 
-The persistence layer itself is tested in `crow-memory`. Integration and end-to-end tiers are opt-in:
+The persistence layer itself is tested in `crow-cli/tests/unit/test_db.py`. Integration and end-to-end tiers are opt-in:
 
 ```bash
 uv run --project crow-cli pytest crow-cli/tests --run-integration   # spawn the agent
@@ -190,9 +189,6 @@ uv run --project crow-cli pytest crow-cli/tests --run-e2e           # live LLM c
 ```
 crow-cli/               the agent — ACP server, ReAct loop, CLI
 crow-mcp/               built-in MCP tool server
-crow-memory/            persistence + memory service (LanceDB, ColBERT/ColPali)
-crow-task-mcp/          task-list MCP server for delegation (being reworked)
-crow-orchestrator-mcp/  orchestration MCP server for delegation (being reworked)
 ```
 
 ## License

{crow_cli-0.1.30 → crow_cli-0.1.32}/README.md

@@ -10,12 +10,12 @@
 
 `crow-cli` is an [Agent Client Protocol (ACP)](https://agentclientprotocol.com/) coding agent that runs in your terminal and inside ACP-compatible editors. It reads and edits code, runs shell commands, searches the web, and remembers your work across sessions.
 
-Most agent toolkits treat persistence as an afterthought. crow-cli treats it as the point: every session lives in a dedicated memory service ([crow-memory](#crow-memory--persistence--memory-api)) built on [LanceDB](https://lancedb.github.io/lancedb/) with ColBERT and ColPali embeddings, so agents recall past conversations semantically and can delegate work to one another. Sessions get memorable coolname ids (like `taupe-squirrel-of-splendid-potency`) you can resume or read from any other agent.
+Most agent toolkits treat persistence as an afterthought. crow-cli treats it as the point: every session lives in a local sqlite database (`~/.agents/crow/crow.db`) with FTS5 full-text search, so agents recall past conversations and can delegate work to one another. Images are stored as files next to the database and hydrated only when sent to the LLM. Sessions get memorable coolname ids (like `taupe-squirrel-of-splendid-potency`) you can resume or read from any other agent.
 
 ## Requirements
 
 - Python 3.14+, managed with [uv](https://docs.astral.sh/uv/)
-- Docker, for the crow-memory and SearXNG services
+- Docker, for the SearXNG service
 - An API key for an OpenAI-compatible LLM provider (OpenRouter, OpenAI, your own endpoint, …)
 
 | Platform | Notes |
@@ -37,11 +37,11 @@ uv tool install crow-cli --python 3.14      # or run without installing: uvx cro
 Initialize your configuration and start the backing services:
 
 ```bash
-crow-cli init                          # scaffolds ~/.crow (config.yaml, .env, docker-compose)
-cd ~/.crow && docker compose up -d     # starts crow-memory + SearXNG
+crow-cli init                          # scaffolds ~/.agents/crow (config.yaml, .env, docker-compose)
+cd ~/.agents/crow && docker compose up -d     # starts SearXNG
 ```
 
-`crow-cli init` walks you through provider and model selection and writes your secrets to `~/.crow/.env`, referenced from the config as `${VAR}`.
+`crow-cli init` walks you through provider and model selection and writes your secrets to `~/.agents/crow/.env`, referenced from the config as `${VAR}`.
 
 ## Quick start
 
@@ -89,17 +89,17 @@ crow-cli is a monorepo. The pieces:
 
 ### crow-cli — the agent
 
-The ACP-native agent: a streaming ReAct loop with tool calling, cancellation, conversation compaction, and multimodal input. Provider and model configuration lives in `~/.crow/config.yaml`.
+The ACP-native agent: a streaming ReAct loop with tool calling, cancellation, conversation compaction, and multimodal input. Provider and model configuration lives in `~/.agents/crow/config.yaml`.
 
-### crow-memory — persistence + memory API
+### Persistence — sqlite memory
 
-A standalone service — a LanceDB store with ColBERT (text) and ColPali (image) multivector embeddings — that the agent talks to over HTTP. It backs both session persistence and a semantic memory API, exposed to agents as three tools:
+Sessions persist to a single sqlite database (`~/.agents/crow/crow.db`, schema v3, WAL mode) with an FTS5 index for BM25 keyword search. Images in messages are written to `~/.agents/crow/images/` and referenced by path; they are hydrated to base64 data URLs only when the conversation is sent to the LLM. The same database backs the memory API, exposed to agents as three tools:
 
 - `list_sessions()` — sessions ordered by recent activity (who's working on what)
 - `query_memory(query)` — find which session discussed something, across all sessions
 - `query_session(session_id)` — read or search within one session (spans all of that session's agents)
 
-This is what makes multi-agent delegation work: launch a worker, then read its thoughts from any other agent. Today crow-memory runs as a Docker container the agent connects to; longer-term it moves toward an always-on daemon, in line with the ACP v2 direction.
+This is what makes multi-agent delegation work: launch a worker, then read its thoughts from any other agent. No service to run — the sqlite file is the integration point.
 
 ### crow-mcp — the tool server
 
@@ -113,7 +113,7 @@ The built-in [MCP](https://modelcontextprotocol.io/) server providing the agent'
 | `capture_webcam` / `read_image_file` | Vision input |
 | `list_sessions` / `query_memory` / `query_session` | Memory (see above) |
 
-**Extensible by design:** register any MCP server in `~/.crow/config.yaml` and its tools appear alongside these automatically.
+**Extensible by design:** register any MCP server in `~/.agents/crow/config.yaml` and its tools appear alongside these automatically.
 
 > ⚠️ **Tool names are not namespaced.** crow-mcp registers its tools as `read`, `edit`, `terminal`, … — not `crow-mcp_read`. When you add your own MCP servers, watch for name collisions.
 
@@ -123,11 +123,11 @@ crow-cli ships a maintained SearXNG configuration (stored as JSON so the agent c
 
 ### Skills
 
-Agents load reusable skills from `~/.crow/skills/` — each a directory with a `SKILL.md` describing when and how to use it. Skill distribution is still being worked out; today skills are local directories.
+Agents load reusable skills from `~/.agents/skills/` — each a directory with a `SKILL.md` describing when and how to use it. Skill distribution is still being worked out; today skills are local directories.
 
 ## Configuration
 
-`~/.crow/config.yaml` holds providers, models, and MCP servers; secrets live in `~/.crow/.env` and are interpolated with `${VAR}`.
+`~/.agents/crow/config.yaml` holds providers, models, and MCP servers; secrets live in `~/.agents/crow/.env` and are interpolated with `${VAR}`.
 
 ```yaml
 providers:
@@ -154,7 +154,7 @@ Run the unit tests — fast and hermetic, no services required (tests that touch
 uv run --project crow-cli pytest crow-cli/tests/unit
 ```
 
-The persistence layer itself is tested in `crow-memory`. Integration and end-to-end tiers are opt-in:
+The persistence layer itself is tested in `crow-cli/tests/unit/test_db.py`. Integration and end-to-end tiers are opt-in:
 
 ```bash
 uv run --project crow-cli pytest crow-cli/tests --run-integration   # spawn the agent
@@ -166,9 +166,6 @@ uv run --project crow-cli pytest crow-cli/tests --run-e2e           # live LLM c
 ```
 crow-cli/               the agent — ACP server, ReAct loop, CLI
 crow-mcp/               built-in MCP tool server
-crow-memory/            persistence + memory service (LanceDB, ColBERT/ColPali)
-crow-task-mcp/          task-list MCP server for delegation (being reworked)
-crow-orchestrator-mcp/  orchestration MCP server for delegation (being reworked)
 ```
 
 ## License

{crow_cli-0.1.30 → crow_cli-0.1.32}/pyproject.toml

@@ -1,35 +1,24 @@
 [project]
 name = "crow-cli"
-version = "0.1.30"
+version = "0.1.32"
 description = "Add your description here"
 readme = "README.md"
 requires-python = ">=3.14"
 dependencies = [
-    "agent-client-protocol>=0.9.0",
+    "agent-client-protocol>=0.12.0,<0.13.0",
     "coolname>=2.2.0",
-    "crow-memory>=0.1.30",
     "directory-tree>=1.0.0",
+    "docker>=7.2.0",
     "fastmcp>=2.14.5",
     "httpx>=0.28.1",
-    "ipython>=9.10.0",
     "jinja2>=3.1.6",
-    "json-schema-to-pydantic>=0.4.9",
     "openai>=2.21.0",
     "pyyaml>=6.0.3",
     "rich>=14.3.3",
+    "sqlalchemy>=2.0.51",
     "typer>=0.24.1",
 ]
 
-[tool.uv.sources]
-crow-memory = { path = "../crow-memory", editable = true }
-torch = [{ index = "pytorch-cpu" }]
-torchvision = [{ index = "pytorch-cpu" }]
-
-[[tool.uv.index]]
-name = "pytorch-cpu"
-url = "https://download.pytorch.org/whl/cpu"
-explicit = true
-
 
 [project.urls]
 Homepage = "https://github.com/crow-cli/crow-cli"
@@ -49,6 +38,7 @@ packages = ["src/crow_cli"]
 
 [dependency-groups]
 dev = [
+    "ipython>=9.10.0",
     "pytest>=9.0.2",
     "pytest-asyncio>=0.23.0",
     "pyinstaller>=6.0.0",

{crow_cli-0.1.30 → crow_cli-0.1.32}/src/crow_cli/agent/compact.py

@@ -146,7 +146,7 @@ async def compact(
     # 5. Create new agent record: same session_id, next agent_idx
     new_agent_idx = original_agent_idx + 1
     new_agent_id = f"{original_session_id}-{new_agent_idx}"
-    new_session = make_agent_session(
+    new_session = await make_agent_session(
         config,
         session.tools,
         session.model_identifier if session.model_identifier else "",
@@ -157,7 +157,7 @@ async def compact(
 
     last_msgs = last_messages(session)
     new_agent_prompt = f"{summary}\n\nLast messages:\n\n{last_msgs}"
-    new_session.add_message(
+    await new_session.add_message(
         {"role": "user", "content": new_agent_prompt},
     )
 

{crow_cli-0.1.30 → crow_cli-0.1.32}/src/crow_cli/agent/configure.py

@@ -2,7 +2,7 @@
 Configuration management for crow-cli.
 
 Default config files are Python string constants imported from defaults.
-On first access to ~/.crow, defaults are written to disk if nothing exists.
+On first access to ~/.agents/crow, defaults are written to disk if nothing exists.
 The user's config.yaml is read from disk for actual runtime config.
 """
 
@@ -25,13 +25,19 @@ from crow_cli.agent.logger import setup_logger
 
 ENV_PATTERN = re.compile(r"\$\{([^}]+)\}")
 
-CROW_DIR = Path.home() / ".crow"
+AGENTS_DIR = Path.home() / ".agents"
+# Config lives under ~/.agents/crow; the shared workspaces (skills, notes,
+# global AGENTS.md) are SIBLINGS of the config dir, directly under ~/.agents.
+DEFAULT_CONFIG_DIR = AGENTS_DIR / "crow"
+SKILLS_DIR = AGENTS_DIR / "skills"
+NOTES_DIR = AGENTS_DIR / "notes"
+GLOBAL_AGENTS_MD = AGENTS_DIR / "AGENTS.md"
 
 
 def get_default_config_dir(config_dir: Path | str | None = None) -> Path:
-    """Return the config directory. If config_dir is given, use it. Otherwise ~/.crow."""
+    """Return the config directory. If config_dir is given, use it. Otherwise ~/.agents/crow."""
     if config_dir is None:
-        return CROW_DIR
+        return DEFAULT_CONFIG_DIR
     return Path(config_dir).resolve()
 
 
@@ -56,14 +62,29 @@ def _write_defaults_if_missing(config_dir: Path) -> None:
             target.write_text(content)
 
 
-def resolve_env_vars(value: Any) -> Any:
-    """Recursively replace ${VAR} with environment variable values."""
+def resolve_env_vars(value: Any, missing: set[str] | None = None) -> Any:
+    """Recursively replace ${VAR} with environment variable values.
+
+    Unset variables expand to "" (parity with the Rust CLI), but their names
+    are collected in `missing` so the caller can warn instead of failing later
+    with an opaque provider error.
+    """
     if isinstance(value, str):
-        return ENV_PATTERN.sub(lambda m: os.getenv(m.group(1), ""), value)
+
+        def _sub(m: re.Match) -> str:
+            name = m.group(1)
+            val = os.getenv(name)
+            if val is None:
+                if missing is not None:
+                    missing.add(name)
+                return ""
+            return val
+
+        return ENV_PATTERN.sub(_sub, value)
     elif isinstance(value, dict):
-        return {k: resolve_env_vars(v) for k, v in value.items()}
+        return {k: resolve_env_vars(v, missing) for k, v in value.items()}
     elif isinstance(value, list):
-        return [resolve_env_vars(v) for v in value]
+        return [resolve_env_vars(v, missing) for v in value]
     return value
 
 
@@ -79,6 +100,12 @@ class LLModel:
     name: str
     provider_name: str
     model_id: str
+    # None = unknown → assume capable of everything (permissive default).
+    # A set restricts: e.g. set() is text-only, {"vision"} handles images.
+    capabilities: set[str] | None = None
+    # Ordered fallback chain (model NAMES from this config) used when this
+    # model cannot handle the modalities present in the conversation.
+    fallbacks: list[str] = field(default_factory=list)
 
 
 @dataclass
@@ -92,6 +119,7 @@ class Config:
     config_dir: Path
     llm: LLMConfig = field(default_factory=LLMConfig)
     memory_path: str = ""
+    skills_dir: str = str(SKILLS_DIR)
     mcp_servers: dict[str, Any] = field(default_factory=dict)
     max_retries_per_step: int = 3
     MAX_COMPACT_TOKENS: int = 190000
@@ -128,21 +156,29 @@ class Config:
         if env_file.exists():
             load_dotenv(env_file)
 
+
+
         # If no config.yaml, return a bare Config
         config_file = target_dir / "config.yaml"
         if not config_file.exists():
             _logger.info("No config.yaml found, returning bare Config")
             return cls(
                 config_dir=target_dir,
-                memory_path=str(Path.home() / ".crow" / "memory.lance"),
+                memory_path=str(DEFAULT_CONFIG_DIR / "crow.db"),
             )
 
         with open(config_file) as f:
             raw = yaml.safe_load(f) or {}
 
         _logger.info("RAW config.yaml mcpServers: %s", raw.get("mcpServers", {}))
-        parsed = resolve_env_vars(raw)
+        missing_vars: set[str] = set()
+        parsed = resolve_env_vars(raw, missing_vars)
         _logger.info("PARSED config.yaml mcpServers after resolve_env_vars: %s", parsed.get("mcpServers", {}))
+        if missing_vars:
+            _logger.warning(
+                "config.yaml references unset environment variables (expanded to empty): %s",
+                ", ".join(f"${{{v}}}" for v in sorted(missing_vars)),
+            )
 
         # Parse providers
         llm = LLMConfig()
@@ -155,14 +191,19 @@ class Config:
 
         # Parse models
         for name, data in parsed.get("models", {}).items():
+            raw_caps = data.get("capabilities")
+            capabilities = set(raw_caps) if raw_caps is not None else None
             llm.models[name] = LLModel(
                 name=name,
                 provider_name=data.get("provider", ""),
                 model_id=data.get("model", ""),
+                capabilities=capabilities,
+                fallbacks=list(data.get("fallbacks") or []),
             )
 
         # Parse overrides
-        memory_path = os.path.expanduser(parsed.get("memory_path") or "~/.crow/memory.lance")
+        memory_path = os.path.expanduser(parsed.get("memory_path") or "~/.agents/crow/crow.db")
+        skills_dir = os.path.expanduser(parsed.get("skills_dir") or str(SKILLS_DIR))
         overrides = {}
         for key, typ in (
             ("max_retries_per_step", int),
@@ -179,12 +220,13 @@ class Config:
         _logger.info("FINAL mcp_servers stored in Config: %s", mcp_servers)
         system_prompt_path = None
         if "system_prompt_path" in parsed:
-            system_prompt_path = Path(parsed["system_prompt_path"])
+            system_prompt_path = Path(os.path.expanduser(parsed["system_prompt_path"]))
 
         return cls(
             config_dir=target_dir,
             llm=llm,
             memory_path=memory_path,
+            skills_dir=skills_dir,
             mcp_servers=mcp_servers,
             system_prompt_path=system_prompt_path,
             **overrides,