Adapters¶

BaseExecutor¶

from aegis.adapters.base import BaseExecutor

Abstract base class. Only execute() is required.

Method	Required	Description
`execute(action)`	Yes	Execute a single action
`verify(action, result)`	No	Post-execution verification
`setup()`	No	Initialize resources
`teardown()`	No	Clean up resources

PlaywrightExecutor¶

from aegis.adapters.playwright import PlaywrightExecutor

executor = PlaywrightExecutor(headless=True, browser_type="chromium")

Action Type	Params	Description
`navigate`	`{"url": str}`	Go to a URL
`click`	`{"selector": str}`	Click an element
`fill`	`{"selector": str, "value": str}`	Fill an input field
`read`	`{"selector": str}`	Extract text content
`screenshot`	`{"path": str}`	Take a screenshot

Requires: pip install 'agent-aegis[playwright]'

LangChainExecutor¶

from aegis.adapters.langchain import LangChainExecutor

executor = LangChainExecutor(tools=[...])
executor.register_tool(another_tool)

Maps Action.type to LangChain tool names.

Requires: pip install 'agent-aegis[langchain]'

AegisTool (LangChain)¶

from aegis.adapters.langchain import AegisTool

tool = AegisTool.from_runtime(
    runtime=runtime,
    name="tool_name",
    description="description",
    action_type="action_type",
    action_target="target",
)

Creates a LangChain StructuredTool that routes through Aegis.

AegisCrewAITool¶

from aegis.adapters.crewai import AegisCrewAITool

tool = AegisCrewAITool(
    runtime=runtime,
    name="tool_name",
    description="description",
    action_type="action_type",
    fn=my_function,
)

Requires: pip install 'agent-aegis[crewai]'

HttpxExecutor¶

from aegis.adapters.httpx_adapter import HttpxExecutor

executor = HttpxExecutor(
    base_url="https://api.example.com",
    default_headers={"Authorization": "Bearer ..."},
    timeout=30.0,
)

Action Type	HTTP Method
`get`	GET
`post`	POST
`put`	PUT
`patch`	PATCH
`delete`	DELETE
`head`	HEAD
`options`	OPTIONS

Supported params keys: json, data, headers, query, timeout.

Requires: pip install 'agent-aegis[httpx]'

@governed_tool (OpenAI Agents SDK)¶

from aegis.adapters.openai_agents import governed_tool

@governed_tool(runtime=runtime, action_type="write", action_target="crm")
async def my_tool(query: str) -> str:
    ...

Requires: pip install 'agent-aegis[openai-agents]'

govern_tool_call (Anthropic Claude)¶

from aegis.adapters.anthropic import govern_tool_call

result = await govern_tool_call(
    runtime=runtime,
    tool_name="update_contact",
    tool_input={"name": "Alice"},
    target="crm",
)

Requires: pip install 'agent-aegis[anthropic]'

govern_mcp_tool_call (MCP)¶

from aegis.adapters.mcp import govern_mcp_tool_call

result = await govern_mcp_tool_call(
    runtime=runtime,
    tool_name="read_file",
    arguments={"path": "/data.csv"},
    server_name="filesystem",
)

Parameter	Type	Description
`runtime`	`Runtime`	Aegis runtime instance
`tool_name`	`str`	MCP tool name (becomes `Action.type`)
`arguments`	`dict`	Tool arguments (becomes `Action.params`)
`server_name`	`str`	MCP server name (becomes `Action.target`)
`description`	`str`	Optional description

AegisMCPToolFilter¶

from aegis.adapters.mcp import AegisMCPToolFilter

tool_filter = AegisMCPToolFilter(runtime=runtime)

# Dry-run check
result = await tool_filter.check(server="filesystem", tool="delete_file")

# Full governance pipeline
result = await tool_filter.call_tool(
    server="filesystem", tool="read_file",
    arguments={"path": "/data.csv"},
)

Method	Description
`check(server, tool, arguments)`	Dry-run policy check
`call_tool(server, tool, arguments)`	Full governance pipeline