Cortex-Inara/cortex/tools/__init__.py

"""
Orchestrator tool registry.

Declarations live in each domain module alongside their callables.
This file assembles them into the unified registry used by both engines.

To add a new tool:
  1. Implement it in tools/<domain>.py — add the async callable + append to DECLARATIONS
  2. Import the callable here and add it to _CALLABLES
  3. If admin-only, add it to TOOL_ROLES; if confirmation needed, add to CONFIRM_REQUIRED

IMPORTANT: These tools are separate from the ae_* MCP tools used by the fleet agents.
           Do not modify the ae_* MCP server to support orchestrator needs.
"""

from google.genai import types

# ── Callable imports ──────────────────────────────────────────────────────────

from tools.web import search as _web_search, http_fetch as _http_fetch, web_read as _web_read, http_post as _http_post
from tools.ae_knowledge import (
    journal_list         as _ae_journal_list,
    journal_search       as _ae_journal_search,
    journal_entry_read   as _ae_journal_entry_read,
    journal_entries_list as _ae_journal_entries_list,
    journal_entry_create as _ae_journal_entry_create,
    journal_entry_update as _ae_journal_entry_update,
    journal_entry_disable as _ae_journal_entry_disable,
    journal_entry_append as _ae_journal_entry_append,
    journal_entry_prepend as _ae_journal_entry_prepend,
)
from tools.ae_tasks import task_list as _ae_task_list
from tools.files import (
    project_file_read  as _project_file_read,
    project_file_list  as _project_file_list,
    file_stat          as _file_stat,
    file_grep          as _file_grep,
    file_diff          as _file_diff,
    file_syntax_check  as _file_syntax_check,
    file_read          as _file_read,
    file_list          as _file_list,
    file_write         as _file_write,
    session_read       as _session_read,
    session_search     as _session_search,
)
from tools.system import (
    shell_exec      as _shell_exec,
    claude_allow_dir as _claude_allow_dir,
    cortex_restart  as _cortex_restart,
    cortex_logs     as _cortex_logs,
    cortex_status   as _cortex_status,
    cortex_update   as _cortex_update,
)
from tools.tasks import (
    task_list     as _task_list,
    task_create   as _task_create,
    task_update   as _task_update,
    task_complete as _task_complete,
)
from tools.cron import (
    cron_list   as _cron_list,
    cron_add    as _cron_add,
    cron_remove as _cron_remove,
    cron_toggle as _cron_toggle,
)
from tools.reminders import (
    reminders_add    as _reminders_add,
    reminders_list   as _reminders_list,
    reminders_remove as _reminders_remove,
    reminders_clear  as _reminders_clear,
)
from tools.scratch import (
    scratch_read   as _scratch_read,
    scratch_write  as _scratch_write,
    scratch_append as _scratch_append,
    scratch_clear  as _scratch_clear,
)
from tools.notify import nc_talk_send as _nc_talk_send, email_send as _email_send, web_push as _web_push, nc_talk_history as _nc_talk_history
from tools.agent_notes import (
    agent_notes_read   as _agent_notes_read,
    agent_notes_write  as _agent_notes_write,
    agent_notes_append as _agent_notes_append,
    agent_notes_clear  as _agent_notes_clear,
)
from tools.agents import spawn_agent as _spawn_agent
from tools.homeassistant import (
    ha_get_state    as _ha_get_state,
    ha_get_states   as _ha_get_states,
    ha_call_service as _ha_call_service,
)

# ── Declaration imports ───────────────────────────────────────────────────────

import tools.web          as _mod_web
import tools.ae_knowledge as _mod_ae_knowledge
import tools.ae_tasks     as _mod_ae_tasks
import tools.files        as _mod_files
import tools.system       as _mod_system
import tools.tasks        as _mod_tasks
import tools.cron         as _mod_cron
import tools.reminders    as _mod_reminders
import tools.scratch      as _mod_scratch
import tools.notify       as _mod_notify
import tools.agent_notes  as _mod_agent_notes
import tools.agents        as _mod_agents
import tools.homeassistant as _mod_homeassistant

# ── Tool categories — used by the Model Registry UI for grouped checkboxes ───

TOOL_CATEGORIES: dict[str, list[str]] = {
    "Web":              ["web_search", "http_fetch", "web_read", "http_post"],
    "Project Files":    ["project_file_read", "project_file_list", "file_stat", "file_grep", "file_diff", "file_syntax_check"],
    "System Files":     ["file_read", "file_list", "file_write", "session_read", "session_search"],
    "Shell":            ["shell_exec", "claude_allow_dir"],
    "System":           ["cortex_restart", "cortex_logs", "cortex_status", "cortex_update"],
    "Tasks":            ["task_list", "task_create", "task_update", "task_complete"],
    "Cron":             ["cron_list", "cron_add", "cron_remove", "cron_toggle"],
    "Reminders":        ["reminders_add", "reminders_list", "reminders_remove", "reminders_clear"],
    "Scratchpad":       ["scratch_read", "scratch_write", "scratch_append", "scratch_clear"],
    "Notifications":    ["web_push", "email_send", "nc_talk_send", "nc_talk_history"],
    "Aether Journals":  [
        "ae_journal_list", "ae_journal_search",
        "ae_journal_entries_list", "ae_journal_entry_read",
        "ae_journal_entry_create", "ae_journal_entry_update",
        "ae_journal_entry_disable", "ae_journal_entry_append",
        "ae_journal_entry_prepend",
    ],
    "Aether Tasks":     ["ae_task_list"],
    "Agent Notes":      ["agent_notes_read", "agent_notes_write", "agent_notes_append", "agent_notes_clear"],
    "Agents":           ["spawn_agent"],
    "Home Assistant":   ["ha_get_state", "ha_get_states", "ha_call_service"],
}

# ── Callable registry ─────────────────────────────────────────────────────────

_CALLABLES: dict[str, callable] = {
    "web_search":                _web_search,
    "http_fetch":                _http_fetch,
    "web_read":                  _web_read,
    "http_post":                 _http_post,
    "ae_journal_list":           _ae_journal_list,
    "ae_journal_search":         _ae_journal_search,
    "ae_journal_entry_read":     _ae_journal_entry_read,
    "ae_journal_entries_list":   _ae_journal_entries_list,
    "ae_journal_entry_create":   _ae_journal_entry_create,
    "ae_journal_entry_update":   _ae_journal_entry_update,
    "ae_journal_entry_disable":  _ae_journal_entry_disable,
    "ae_journal_entry_append":   _ae_journal_entry_append,
    "ae_journal_entry_prepend":  _ae_journal_entry_prepend,
    "ae_task_list":              _ae_task_list,
    "project_file_read":         _project_file_read,
    "project_file_list":         _project_file_list,
    "file_stat":                 _file_stat,
    "file_grep":                 _file_grep,
    "file_diff":                 _file_diff,
    "file_syntax_check":         _file_syntax_check,
    "file_read":                 _file_read,
    "file_list":                 _file_list,
    "file_write":                _file_write,
    "session_read":              _session_read,
    "session_search":            _session_search,
    "shell_exec":                _shell_exec,
    "claude_allow_dir":          _claude_allow_dir,
    "cortex_restart":            _cortex_restart,
    "cortex_logs":               _cortex_logs,
    "cortex_status":             _cortex_status,
    "cortex_update":             _cortex_update,
    "task_list":                 _task_list,
    "task_create":               _task_create,
    "task_update":               _task_update,
    "task_complete":             _task_complete,
    "cron_list":                 _cron_list,
    "cron_add":                  _cron_add,
    "cron_remove":               _cron_remove,
    "cron_toggle":               _cron_toggle,
    "reminders_add":             _reminders_add,
    "reminders_list":            _reminders_list,
    "reminders_remove":          _reminders_remove,
    "reminders_clear":           _reminders_clear,
    "scratch_read":              _scratch_read,
    "scratch_write":             _scratch_write,
    "scratch_append":            _scratch_append,
    "scratch_clear":             _scratch_clear,
    "email_send":                _email_send,
    "nc_talk_send":              _nc_talk_send,
    "web_push":                  _web_push,
    "nc_talk_history":           _nc_talk_history,
    "agent_notes_read":          _agent_notes_read,
    "agent_notes_write":         _agent_notes_write,
    "agent_notes_append":        _agent_notes_append,
    "agent_notes_clear":         _agent_notes_clear,
    "spawn_agent":               _spawn_agent,
    "ha_get_state":              _ha_get_state,
    "ha_get_states":             _ha_get_states,
    "ha_call_service":           _ha_call_service,
}

# ── Role-based access control ─────────────────────────────────────────────────

# Minimum role required to use each tool. Unlisted tools default to "user".
TOOL_ROLES: dict[str, str] = {
    "shell_exec":       "admin",
    "claude_allow_dir": "admin",
    "cortex_restart":   "admin",
    "cortex_logs":      "admin",
    "cortex_status":    "admin",
    "cortex_update":    "admin",
    "file_read":        "admin",
    "file_list":        "admin",
    "file_write":       "admin",
    "ae_task_list":     "admin",
    "spawn_agent":      "admin",
    "email_send":       "admin",
    "nc_talk_send":     "admin",
    "http_post":        "admin",
    "nc_talk_history":  "admin",
    "ha_call_service":  "admin",
}

# Tools that require explicit user confirmation before executing.
CONFIRM_REQUIRED: set[str] = {
    "cortex_restart",
    "cortex_update",
    "file_write",
    "shell_exec",
    "cron_remove",
    "reminders_clear",
    "http_post",
    "ha_call_service",
}

# Security risk ratings — informational for now; will drive auto-allow tiers later.
# Unlisted tools default to "medium".
#
#   low    — read-only, sandboxed, no external side effects
#   medium — writes to local/controlled data, or reads beyond project scope,
#            or sends notifications to the same user
#   high   — affects external systems, physical devices, other users,
#            or the host process/filesystem in ways that are hard to reverse
TOOL_RISK: dict[str, str] = {
    # Web — read-only fetches are low; posting to external services is high
    "web_search":                "low",
    "http_fetch":                "low",
    "web_read":                  "low",
    "http_post":                 "high",

    # Project Files — all read-only and project-sandboxed
    "project_file_read":         "low",
    "project_file_list":         "low",
    "file_stat":                 "low",
    "file_grep":                 "low",
    "file_diff":                 "low",
    "file_syntax_check":         "low",

    # System Files — reads beyond project scope are medium; writes are high
    "file_read":                 "medium",
    "file_list":                 "medium",
    "file_write":                "high",
    "session_read":              "low",
    "session_search":            "low",

    # Shell — arbitrary execution and permission changes are high
    "shell_exec":                "high",
    "claude_allow_dir":          "high",

    # System — read-only status is low; restart/update affect the live service
    "cortex_logs":               "low",
    "cortex_status":             "low",
    "cortex_restart":            "high",
    "cortex_update":             "high",

    # Tasks — local persona data, all reversible
    "task_list":                 "low",
    "task_create":               "low",
    "task_update":               "low",
    "task_complete":             "low",

    # Cron — list is low; add/remove/toggle affect scheduled behavior
    "cron_list":                 "low",
    "cron_add":                  "medium",
    "cron_remove":               "medium",
    "cron_toggle":               "medium",

    # Reminders — single-item ops are low; clear-all is medium
    "reminders_add":             "low",
    "reminders_list":            "low",
    "reminders_remove":          "low",
    "reminders_clear":           "medium",

    # Scratchpad — local persona file, ephemeral by design
    "scratch_read":              "low",
    "scratch_write":             "low",
    "scratch_append":            "low",
    "scratch_clear":             "low",

    # Notifications — push to same user is medium; external messages are high
    "web_push":                  "medium",
    "nc_talk_send":              "high",
    "nc_talk_history":           "low",
    "email_send":                "high",

    # Aether Journals — reads are low; writes to external DB are medium
    "ae_journal_list":           "low",
    "ae_journal_search":         "low",
    "ae_journal_entries_list":   "low",
    "ae_journal_entry_read":     "low",
    "ae_journal_entry_create":   "medium",
    "ae_journal_entry_update":   "medium",
    "ae_journal_entry_disable":  "medium",
    "ae_journal_entry_append":   "medium",
    "ae_journal_entry_prepend":  "medium",

    # Aether Tasks
    "ae_task_list":              "low",

    # Agent Notes — local persona file
    "agent_notes_read":          "low",
    "agent_notes_write":         "low",
    "agent_notes_append":        "low",
    "agent_notes_clear":         "low",

    # Agents — spawning a subprocess with broad permissions is high
    "spawn_agent":               "high",

    # Home Assistant — reads are low; controlling physical devices is high
    "ha_get_state":              "low",
    "ha_get_states":             "low",
    "ha_call_service":           "high",
}

_RISK_RANK: dict[str, int] = {"low": 0, "medium": 1, "high": 2}

_ROLE_RANK: dict[str, int] = {"user": 0, "admin": 1}


def _role_allowed(tool_name: str, role: str) -> bool:
    required = TOOL_ROLES.get(tool_name, "user")
    return _ROLE_RANK.get(role, 0) >= _ROLE_RANK.get(required, 0)


# ── Declaration assembly ──────────────────────────────────────────────────────

_ALL_DECLARATIONS: list[types.FunctionDeclaration] = (
    _mod_web.DECLARATIONS
    + _mod_files.DECLARATIONS
    + _mod_system.DECLARATIONS
    + _mod_tasks.DECLARATIONS
    + _mod_cron.DECLARATIONS
    + _mod_reminders.DECLARATIONS
    + _mod_scratch.DECLARATIONS
    + _mod_notify.DECLARATIONS
    + _mod_ae_knowledge.DECLARATIONS
    + _mod_ae_tasks.DECLARATIONS
    + _mod_agent_notes.DECLARATIONS
    + _mod_agents.DECLARATIONS
    + _mod_homeassistant.DECLARATIONS
)

# Full Gemini Tool object (all tools — use get_tools_for_role() in production)
TOOL_DECLARATIONS = [types.Tool(function_declarations=_ALL_DECLARATIONS)]


# ── Tool dispatch ─────────────────────────────────────────────────────────────

async def call_tool(name: str, args: dict, callables: dict | None = None) -> str:
    """Dispatch a tool call by name. Returns result as a string.

    Pass `callables` (from get_tools_for_role) to enforce role restrictions.
    Falls back to the full _CALLABLES dict if omitted.

    Every call is recorded to the tool audit log (tool_audit.py).
    """
    import asyncio
    import tool_audit
    from persona import get_user

    user = get_user() or "unknown"
    dispatch = callables if callables is not None else _CALLABLES
    fn = dispatch.get(name)

    if fn is None:
        asyncio.create_task(tool_audit.record(user, name, args, "denied"))
        return f"Tool not available or access denied: {name}"

    try:
        result = await fn(**args)
        asyncio.create_task(tool_audit.record(user, name, args, "ok", result))
        return result
    except Exception as e:
        asyncio.create_task(tool_audit.record(user, name, args, "error", str(e)))
        raise


# ── OpenAI JSON Schema conversion ────────────────────────────────────────────

_GEMINI_TYPE_TO_JSON = {
    "OBJECT":  "object",
    "STRING":  "string",
    "INTEGER": "integer",
    "NUMBER":  "number",
    "BOOLEAN": "boolean",
    "ARRAY":   "array",
}


def _schema_to_json(schema) -> dict:
    """Recursively convert a Gemini types.Schema to a JSON Schema dict."""
    type_name = getattr(getattr(schema, "type", None), "name", "STRING")
    result: dict = {"type": _GEMINI_TYPE_TO_JSON.get(type_name, "string")}

    if getattr(schema, "description", None):
        result["description"] = schema.description

    props = getattr(schema, "properties", None) or {}
    if result["type"] == "object":
        result["properties"] = {k: _schema_to_json(v) for k, v in props.items()}

    req = getattr(schema, "required", None)
    if req:
        result["required"] = list(req)

    return result


def _build_openai_tools() -> list[dict]:
    out = []
    for decl in _ALL_DECLARATIONS:
        params = (
            _schema_to_json(decl.parameters)
            if decl.parameters
            else {"type": "object", "properties": {}}
        )
        out.append({
            "type": "function",
            "function": {
                "name": decl.name,
                "description": decl.description or "",
                "parameters": params,
            },
        })
    return out


# OpenAI-format tool list — all tools (use get_openai_tools_for_role() in production)
OPENAI_TOOL_SCHEMAS: list[dict] = _build_openai_tools()


# ── Role-filtered tool access ─────────────────────────────────────────────────

def _apply_risk_policy(
    allowed: set[str],
    max_risk: str | None,
    whitelist: list[str] | None,
    blacklist: list[str] | None,
) -> set[str]:
    """Apply risk-level filtering on top of an already role-gated allowed set.

    Filtering order (each step can only restrict or restore within what the
    role already permits — risk policy can never elevate above role):

      1. max_risk auto-include: keep tools whose risk ≤ max_risk
      2. whitelist union: force-add specific tools (still role-gated)
      3. blacklist subtract: force-remove specific tools

    When max_risk is None, all role-allowed tools remain (no risk filter).
    """
    if max_risk is not None:
        max_rank = _RISK_RANK.get(max_risk, 2)
        auto = {n for n in allowed if _RISK_RANK.get(TOOL_RISK.get(n, "medium"), 1) <= max_rank}
        extra = {n for n in (whitelist or []) if n in allowed}
        allowed = (auto | extra)
    if blacklist:
        allowed -= set(blacklist)
    return allowed


def get_tools_for_role(
    role: str,
    tool_list: list[str] | None = None,
    max_risk: str | None = None,
    whitelist: list[str] | None = None,
    blacklist: list[str] | None = None,
) -> tuple[list, dict]:
    """Return (gemini_tool_declarations, callables_dict) filtered to what the role can use.

    role      — user access level ("user" | "admin"); gates admin-only tools
    tool_list — optional model-level allow-list; intersected so it can only restrict
    max_risk  — auto-include tools at/below this risk level ("low"|"medium"|"high")
    whitelist — force-include specific tools above max_risk (still role-gated)
    blacklist — force-exclude specific tools regardless of max_risk
    """
    allowed = {name for name in _CALLABLES if _role_allowed(name, role)}
    allowed = _apply_risk_policy(allowed, max_risk, whitelist, blacklist)
    if tool_list is not None:
        allowed &= set(tool_list)
    decls = [d for d in _ALL_DECLARATIONS if d.name in allowed]
    callables = {k: v for k, v in _CALLABLES.items() if k in allowed}
    return [types.Tool(function_declarations=decls)], callables


def get_openai_tools_for_role(
    role: str,
    tool_list: list[str] | None = None,
    max_risk: str | None = None,
    whitelist: list[str] | None = None,
    blacklist: list[str] | None = None,
) -> list[dict]:
    """Return OpenAI tool schemas filtered to what the role can use.

    role      — user access level ("user" | "admin")
    tool_list — optional model-level allow-list
    max_risk  — auto-include tools at/below this risk level
    whitelist — force-include specific tools above max_risk
    blacklist — force-exclude specific tools
    """
    allowed = {name for name in _CALLABLES if _role_allowed(name, role)}
    allowed = _apply_risk_policy(allowed, max_risk, whitelist, blacklist)
    if tool_list is not None:
        allowed &= set(tool_list)
    return [t for t in OPENAI_TOOL_SCHEMAS if t["function"]["name"] in allowed]