inspect-swe 0.2.6__tar.gz → 0.2.8__tar.gz

{inspect_swe-0.2.6 → inspect_swe-0.2.8}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: inspect_swe
-Version: 0.2.6
+Version: 0.2.8
 Summary: Software engineering agents for Inspect AI.
 Project-URL: Documentation, https://meridianlabs-ai.github.io/inspect_swe/
 Project-URL: Source Code, https://github.com/meridianlabs-ai/inspect_swe
@@ -10,7 +10,7 @@ License: MIT License
 License-File: LICENSE
 Requires-Python: >=3.10
 Requires-Dist: httpx
-Requires-Dist: inspect-ai>=0.3.126
+Requires-Dist: inspect-ai>=0.3.128
 Requires-Dist: nest-asyncio
 Requires-Dist: platformdirs
 Requires-Dist: pydantic>=2.11.4

{inspect_swe-0.2.6 → inspect_swe-0.2.8}/pyproject.toml

@@ -12,7 +12,7 @@ requires-python = ">=3.10"
 license = { text = "MIT License" }
 dependencies = [
     "httpx",
-    "inspect_ai>=0.3.126",
+    "inspect_ai>=0.3.128",
     "nest_asyncio",
     "platformdirs",
     "pydantic>=2.11.4",

{inspect_swe-0.2.6 → inspect_swe-0.2.8}/src/inspect_swe/_claude_code/claude_code.py

@@ -29,13 +29,16 @@ def claude_code(
     """),
     system_prompt: str | None = None,
     mcp_servers: Sequence[MCPServerConfig] | None = None,
+    allowed_tools: list[str] | None = None,
+    disallowed_tools: list[str] | None = None,
     attempts: int | AgentAttempts = 1,
     model: str | None = None,
     small_model: str | None = None,
+    cwd: str | None = None,
     env: dict[str, str] | None = None,
-    version: Literal["auto", "sandbox", "stable", "latest"] | str = "auto",
     user: str | None = None,
     sandbox: str | None = None,
+    version: Literal["auto", "sandbox", "stable", "latest"] | str = "auto",
 ) -> Agent:
     """Claude Code agent.
 
@@ -43,6 +46,8 @@ def claude_code(
 
     The agent can either use a version of Claude Code installed in the sandbox, or can download a version and install it in the sandbox (see docs on `version` option below for details).
 
+    Use `allowed_tools` and `disallowed_tools` to control access to tools. See [Tools available to Claude](https://docs.anthropic.com/en/docs/claude-code/settings#tools-available-to-claude) for the list of built-in tools and [How to use Allowed Tools in Claude Code](https://www.instructa.ai/blog/claude-code/how-to-use-allowed-tools-in-claude-code) for details on the supported syntax. Note that `allowed_tools` enables you to filter allowed parameter values and `disallowed_tools` enables you to remove tools entirely. In other words, `allowed_tools` is not a complete list of what tools are available but rather just filters on tool parameters---to remove tools you need to explicitly set `disallowed_tools`.
+
     Use the `attempts` option to enable additional submissions if the initial
     submission(s) are incorrect (by default, no additional attempts are permitted).
 
@@ -51,18 +56,21 @@ def claude_code(
         description: Agent description (used in multi-agent systems with `as_tool()` and `handoff()`)
         system_prompt: Additional system prompt to append to default system prompt.
         mcp_servers: MCP servers to make available to the agent.
+        allowed_tools: Parameter filters for built-in tools.
+        disallowed_tools: List of tool names to disallow entirely.
         attempts: Configure agent to make multiple attempts.
         model: Model name to use for Opus and Sonnet calls (defaults to main model for task).
         small_model: Model to use for Haiku calls (defaults to main model for task).
+        cwd: Working directory to run claude code within.
         env: Environment variables to set for claude code.
+        user: User to execute claude code with.
+        sandbox: Optional sandbox environment name.
         version: Version of claude code to use. One of:
             - "auto": Use any available version of claude code in the sandbox, otherwise download the current stable version.
             - "sandbox": Use the version of claude code in the sandbox (raises `RuntimeError` if claude is not available in the sandbox)
             - "stable": Download and use the current stable version of claude code.
             - "latest": Download and use the very latest version of claude code.
             - "x.x.x": Download and use a specific version of claude code.
-        user: User to execute claude code with.
-        sandbox: Optional sandbox environment name.
     """
     # resolve models
     model = f"inspect/{model}" if model is not None else "inspect"
@@ -99,8 +107,19 @@ def claude_code(
                 cmd.extend(["--append-system-prompt", "\n\n".join(system_messages)])
 
             # mcp servers
+            cmd_allowed_tools = allowed_tools or []
             if mcp_servers:
-                cmd.extend(mcp_server_args(mcp_servers))
+                mcp_server_args, mcp_allowed_tools = resolve_mcp_servers(mcp_servers)
+                cmd.extend(mcp_server_args)
+                cmd_allowed_tools.extend(mcp_allowed_tools)
+
+            # add allowed and disallowed tools
+            if len(cmd_allowed_tools) > 0:
+                cmd.append("--allowed-tools")
+                cmd.append(",".join(cmd_allowed_tools))
+            if disallowed_tools is not None and len(disallowed_tools) > 0:
+                cmd.append("--disallowed-tools")
+                cmd.append(",".join(disallowed_tools))
 
             # user prompt
             prompt = "\n\n".join(
@@ -123,6 +142,7 @@ def claude_code(
                 # run agent
                 result = await sbox.exec(
                     cmd=agent_cmd,
+                    cwd=cwd,
                     env={
                         "ANTHROPIC_BASE_URL": f"http://localhost:{bridge.port}",
                         "ANTHROPIC_API_KEY": "sk-ant-api03-DOq5tyLPrk9M4hPE",
@@ -174,7 +194,9 @@ def claude_code(
     return agent_with(execute, name=name, description=description)
 
 
-def mcp_server_args(mcp_servers: Sequence[MCPServerConfig]) -> list[str]:
+def resolve_mcp_servers(
+    mcp_servers: Sequence[MCPServerConfig],
+) -> tuple[list[str], list[str]]:
     # build servers and allowed tools
     mcp_servers_json: dict[str, dict[str, Any]] = {}
     allowed_tools: list[str] = []
@@ -194,14 +216,11 @@ def mcp_server_args(mcp_servers: Sequence[MCPServerConfig]) -> list[str]:
             )
 
     # map to cli args
-    cmds: list[str] = []
+    mcp_config_cmds: list[str] = []
     if len(mcp_servers_json) > 0:
-        cmds.append("--mcp-config")
-        cmds.append(
+        mcp_config_cmds.append("--mcp-config")
+        mcp_config_cmds.append(
             to_json({"mcpServers": mcp_servers_json}, exclude_none=True).decode()
         )
-    if len(allowed_tools):
-        cmds.append("--allowed-tools")
-        cmds.append(",".join(allowed_tools))
 
-    return cmds
+    return mcp_config_cmds, allowed_tools

{inspect_swe-0.2.6 → inspect_swe-0.2.8}/src/inspect_swe/_version.py

@@ -28,7 +28,7 @@ version_tuple: VERSION_TUPLE
 commit_id: COMMIT_ID
 __commit_id__: COMMIT_ID
 
-__version__ = version = '0.2.6'
-__version_tuple__ = version_tuple = (0, 2, 6)
+__version__ = version = '0.2.8'
+__version_tuple__ = version_tuple = (0, 2, 8)
 
 __commit_id__ = commit_id = None