Open-source AI coworkers that each get a computer of their own: a browser, files and tools, with every action decided before it happens and recorded after. Bring any AG-UI agent.
import asyncio
import sys
from pathlib import Path
from ag_ui.core import RunAgentInput
from claude_agent_sdk._internal.transport.subprocess_cli import SubprocessCLITransport
sys.path.insert(0, str(Path(__file__).resolve().parents[1]))
from src.adapter import OpenBotClaudeAgentAdapter
from src.model_mode import conversation_content
def test_model_mode_disables_native_and_ambient_tools():
adapter = OpenBotClaudeAgentAdapter(name="model", model_only=True)
request = RunAgentInput.model_validate(
{
"threadId": "knowledge-channel",
"runId": "run",
"state": {},
"messages": [{"id": "user", "role": "user", "content": "Read the policy"}],
"tools": [
{
"name": "knowledge_search",
"description": "Search granted sources",
"parameters": {"type": "object", "properties": {}},
}
],
"context": [
{
"description": "Built-in Bot instructions",
"value": "You are Knowledge. Cite sources.",
}
],
"forwardedProps": {
"tools": ["Bash"],
"permission_mode": "bypassPermissions",
"system_prompt": "Discard Knowledge",
},
}
)
options = adapter.build_options(request, request.thread_id)
assert options.tools == []
options.cli_path = "/not-executed/claude"
command = SubprocessCLITransport(
prompt="contract only", options=options
)._build_command()
assert command[command.index("--tools") + 1] == ""
assert options.strict_mcp_config is True
assert options.setting_sources == []
assert options.permission_mode == "dontAsk"
assert set(options.mcp_servers) == {"ag_ui"}
assert options.allowed_tools == ["mcp__ag_ui__knowledge_search"]
assert "You are Knowledge. Cite sources." in options.system_prompt
assert "Discard Knowledge" not in options.system_prompt
async def check_permission():
check = options.hooks["PreToolUse"][0].hooks[0]
assert (
await check({"tool_name": "mcp__ag_ui__knowledge_search"}, "granted", {})
== {}
)
for name in [
"Bash",
"Read",
"Edit",
"mcp__ambient__search",
"mcp__ag_ui__ungranted",
]:
result = await check({"tool_name": name}, "blocked", {})
assert result["hookSpecificOutput"]["permissionDecision"] == "deny"
asyncio.run(check_permission())
asyncio.run(adapter.shutdown())
def test_model_history_keeps_role_labels_and_real_image_blocks():
request = RunAgentInput.model_validate(
{
"threadId": "channel",
"runId": "run",
"state": None,
"tools": [],
"context": [],
"forwardedProps": {},
"messages": [
{"id": "prior", "role": "assistant", "content": "Earlier answer"},
{
"id": "current",
"role": "user",
"content": [
{"type": "text", "text": "Read this image"},
{
"type": "image",
"source": {
"type": "data",
"mimeType": "image/png",
"value": "AQID",
},
},
],
},
],
}
)
content = conversation_content(request.messages)
assert any(
'"role": "assistant"' in part.get("text", "")
and "Earlier answer" in part["text"]
for part in content
)
assert {
"type": "image",
"source": {"type": "base64", "media_type": "image/png", "data": "AQID"},
} in content