WEBDIENSTE/openswarm
1
0
Fork 0
mirror of https://github.com/openswarm-ai/openswarm.git synced 2026-09-10 11:47:43 +02:00
Code Issues Packages Projects Releases Wiki Activity

[eric] agents: extract web-MCP-need provider decision into tools/web (should_register_web_mcp)

Browse Source
This commit is contained in:
ciregenz
2026-06-23 02:52:55 -07:00
parent 44832b2b13
commit f7f37766b4
3 changed files with 107 additions and 58 deletions
Download Patch File Download Diff File Expand all files Collapse all files
backend/apps/agents/agent_manager.py
+13 -57
View File
@@ -62,6 +62,7 @@ from backend.apps.agents.manager.streaming import thinking as thinking_mod
from backend.apps.agents.manager.streaming import tool_result_hook
from backend.apps.agents.manager.streaming import stop_hook as stop_hook_mod
from backend.apps.agents.manager.prompt.system_prompt import compose_turn_system_prompt
from backend.apps.agents.tools.web import should_register_web_mcp
from backend.apps.agents.manager.permissions import gate_hooks
from backend.apps.agents.manager.view_builder_state import (
view_builder_render_retry_counts,
@@ -552,63 +553,18 @@ class AgentManager:
}
# The CLI's built-in WebSearch/WebFetch wraps Anthropic's
# web_search_20250305. For non-Claude primaries the CLI
# delegates execution back to Anthropic via
# ANTHROPIC_SMALL_FAST_MODEL, needs an Anthropic credential
# or it 401s. We register our DDG-backed MCP only for users
# with no Anthropic path; Anthropic's hosted search is
# higher-quality so we prefer it whenever it's reachable.
# Register the DDG-backed openswarm-web MCP only when the primary has no reliable
# native Anthropic web path (decided in tools/web.py); _m feeds the registration log
# + provider branch just below, so it stays a loop local.
_m = _router_model_id if isinstance(_router_model_id, str) else ""
# When the primary is non-Claude we deliberately don't count
# OpenSwarm Pro as an Anthropic path, using the Pro pool for
# WebSearch on a GPT/Gemini session would drain it for the
# user's Claude turns. The user's GPT/Gemini subscription
# serves their non-Claude turns at zero cost to us.
_primary_is_claude = _m.startswith("cc/") or (
isinstance(_router_model_id, str)
and not _router_model_id.startswith(("cc/", "cx/", "gc/", "ag/", "gemini/"))
and _api_type_for_session == "anthropic"
need_web_mcp = should_register_web_mcp(
model=session.model,
router_model_id=_router_model_id,
api_type=_api_type_for_session,
anthropic_api_key=getattr(global_settings, "anthropic_api_key", None),
connection_mode=getattr(global_settings, "connection_mode", "own_key"),
)
# Custom-provider sessions (Ollama Cloud, Together, Groq, etc.)
# set ANTHROPIC_BASE_URL to 9Router but 9Router has no Claude
# connection unless the user separately set up one. The CLI's
# built-in WebSearch delegates to Anthropic Haiku, which falls
# through 9Router to whichever connection serves anthropic/...
# ids, usually OpenRouter, and 401s. Force the openswarm-web
# MCP to register so WebSearch always cascades through our own
# /api/web/search (Gemini → OpenAI → DuckDuckGo).
_is_custom_session = _api_type_for_session == "custom"
# The built-in WebSearch's aux haiku call only authenticates when it
# reaches an ENTITLED Anthropic endpoint. That's true in exactly two
# cases, mirroring the direct-Anthropic env-branch built further down:
# a direct Anthropic api-route model (base_url = api.anthropic.com
# with the user's key), or OpenSwarm Pro (entitled to the managed pool
# 9Router's anthropic/* resolves to). A SUBSCRIPTION-route Claude
# model (opus-4-8, route=None) routes the haiku call through 9Router
# to the managed pool and 401s for non-Pro users, so a bare key in
# settings is NOT enough; it must be a *-api route model. Everyone
# else registers openswarm-web and cascades through /api/web/search.
from backend.apps.agents.tools.web import anthropic_web_search_is_reliable
from backend.apps.agents.providers.registry import _find_builtin_model as _fbm_web
_web_model_entry = _fbm_web(session.model)
_uses_direct_anthropic_api = (
_web_model_entry is not None
and _web_model_entry.get("route") == "api"
and _web_model_entry.get("api") == "anthropic"
and bool(getattr(global_settings, "anthropic_api_key", None))
)
_has_anthropic_path = (
not _is_custom_session
and _primary_is_claude
and anthropic_web_search_is_reliable(
uses_direct_anthropic_api=_uses_direct_anthropic_api,
is_pro=(getattr(global_settings, "connection_mode", "own_key") in ("openswarm-pro", "free-trial")),
)
)
_need_web_mcp = not _has_anthropic_path
if _need_web_mcp:
if need_web_mcp:
web_mcp_server_path = os.path.join(
os.path.dirname(__file__), "web_mcp_server.py"
)
@@ -701,7 +657,7 @@ class AgentManager:
# WebSearch/WebFetch are guaranteed to fail (no Anthropic
# backend). Suppress them so the model picks our MCP variants
# and doesn't waste a turn on a broken tool.
if _need_web_mcp:
if need_web_mcp:
effective_allowed = [t for t in effective_allowed if t not in ("WebSearch", "WebFetch")]
for _bt in ("WebSearch", "WebFetch"):
if _bt not in effective_disallowed:
@@ -719,7 +675,7 @@ class AgentManager:
# web MCP, AND (b) the user hasn't disabled the policy, matches
# the same gate the MCP allowlist uses, so disabling WebSearch in
# Settings still wins.
_web_tools_available = _need_web_mcp and (
_web_tools_available = need_web_mcp and (
"mcp__openswarm-web__WebSearch" in effective_allowed
or "mcp__openswarm-web__WebFetch" in effective_allowed
)
backend/apps/agents/tools/web.py
+46 -1
View File
@@ -4,9 +4,10 @@ from __future__ import annotations
import html
import re
from typing import Any
from typing import Any, Optional
import httpx
from typeguard import typechecked
from backend.apps.agents.tools.base import BaseTool, ToolContext
from backend.apps.agents.tools.ssrf_guard import SSRFBlocked, safe_fetch
@@ -48,6 +49,50 @@ def anthropic_web_search_is_reliable(*, uses_direct_anthropic_api: bool,
return bool(uses_direct_anthropic_api or is_pro)
@typechecked
def should_register_web_mcp(
*,
model: str,
router_model_id: object,
api_type: Optional[str],
anthropic_api_key: Optional[str],
connection_mode: str,
) -> bool:
"""True when the agent loop must register the DDG-backed openswarm-web MCP because the
primary model has NO reliable native Anthropic web-search path. We prefer Anthropic's
hosted search (return False) whenever it's actually reachable, and cascade through our own
/api/web/search (Gemini -> OpenAI -> DuckDuckGo) otherwise. The three no-path cases:
a non-Claude primary, a custom-provider session (ANTHROPIC_BASE_URL points at 9Router with
no Claude connection), and a subscription-route Claude model on a non-Pro account (the
built-in WebSearch's aux haiku call 401s). Pro pool is deliberately NOT counted for a
non-Claude primary: spending it on WebSearch would drain the user's Claude turns."""
from backend.apps.agents.providers.registry import _find_builtin_model as find_builtin_model
m = router_model_id if isinstance(router_model_id, str) else ""
primary_is_claude = m.startswith("cc/") or (
isinstance(router_model_id, str)
and not router_model_id.startswith(("cc/", "cx/", "gc/", "ag/", "gemini/"))
and api_type == "anthropic"
)
is_custom_session = api_type == "custom"
web_model_entry = find_builtin_model(model)
uses_direct_anthropic_api = (
web_model_entry is not None
and web_model_entry.get("route") == "api"
and web_model_entry.get("api") == "anthropic"
and bool(anthropic_api_key)
)
has_anthropic_path = (
not is_custom_session
and primary_is_claude
and anthropic_web_search_is_reliable(
uses_direct_anthropic_api=uses_direct_anthropic_api,
is_pro=(connection_mode in ("openswarm-pro", "free-trial")),
)
)
return not has_anthropic_path
def _truncate(text: str, limit: int = _MAX_OUTPUT_BYTES) -> str:
if len(text) > limit:
return text[:limit] + "\n... (output truncated)"
backend/tests/test_web_mcp_decision.py
+48
View File
@@ -0,0 +1,48 @@
"""Unit coverage for should_register_web_mcp: the decision of whether to register the
DDG-backed openswarm-web MCP. Security/cost-relevant (it governs whether WebSearch cascades
through our own /api/web/search vs the native Anthropic path), so pin each provider case."""
import pytest
from unittest.mock import patch
from backend.apps.agents.tools.web import should_register_web_mcp
def _call(**kw):
base = dict(model="m", router_model_id="cc/opus", api_type="anthropic",
anthropic_api_key=None, connection_mode="own_key")
base.update(kw)
with patch("backend.apps.agents.providers.registry._find_builtin_model", return_value=None):
return should_register_web_mcp(**base)
def test_custom_session_always_registers():
# ANTHROPIC_BASE_URL points at 9Router with no Claude connection -> native WebSearch 401s.
assert _call(api_type="custom") is True
def test_non_claude_primary_registers():
# A Gemini/GPT primary has no native Anthropic web path; Pro pool is not counted for it.
assert _call(router_model_id="gemini/flash", api_type="google") is True
def test_claude_pro_uses_native_path():
# Claude primary on Pro: the managed pool entitles the built-in WebSearch, so don't register.
assert _call(router_model_id="cc/opus", api_type="anthropic", connection_mode="openswarm-pro") is False
def test_subscription_route_claude_non_pro_registers():
# opus-4-8 on a non-Pro own-key account: the aux haiku call 401s through 9Router, so a bare
# key isn't enough -> fall back to openswarm-web.
assert _call(router_model_id="cc/opus", api_type="anthropic",
connection_mode="own_key", anthropic_api_key="sk-ant-xxx") is True
def test_direct_anthropic_api_route_uses_native_path():
entry = {"route": "api", "api": "anthropic"}
with patch("backend.apps.agents.providers.registry._find_builtin_model", return_value=entry):
out = should_register_web_mcp(
model="claude-x", router_model_id="claude-3-5-api", api_type="anthropic",
anthropic_api_key="sk-ant-xxx", connection_mode="own_key",
)
assert out is False