AI agents call directory_tree to retrieve information from Claude Context without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This tool retrieves filesystem hierarchy information without side effects. It is a read operation analogous to 'ls -R' or similar directory listing commands. While the description is empty (reducing confidence slightly), the tool name and sibling context (semantic code search, file reading) indicate a pure query function. No data is modified, deleted, or executed—only metadata about directory structure is returned.
From the tool's definition Tool name 'directory_tree' indicates a retrieval operation that lists directory structure. Sibling tools like 'list_directory' and 'read_file' on the same server are non-destructive query operations.
Documented attack patterns abuse exactly the kind of access directory_tree gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and Claude Context, and nothing reaches the server without passing your rules. This is the rule we recommend for directory_tree:
{
"version": "1",
"default": "deny",
"tools": {
"directory_tree": {}
}
}directory_tree is read-only, so it stays allowed — but everything else on the server is denied unless you say otherwise.
Free to start. No card required.
directory_tree. It is categorised as a Read tool in the Claude Context MCP Server, which means it retrieves data without modifying state.
Register the Claude Context MCP server in PolicyLayer and add a rule for directory_tree: allow, deny, rate-limit, or require approval. Point your MCP client at the PolicyLayer proxy URL and the rule is enforced on every call, before it reaches Claude Context. Nothing to install.
directory_tree is a Read tool with low risk. Read-only tools are generally safe to allow by default.
Yes. Add a rate_limit block to the directory_tree rule in your PolicyLayer policy. For example, setting max: 10 and window: 60 limits the tool to 10 calls per minute. Rate limits are tracked per agent session and reset automatically.
Set action: deny in the PolicyLayer policy for directory_tree. The AI agent will receive a policy violation error and cannot call the tool. You can also include a reason field to explain why the tool is blocked.
directory_tree is provided by the Claude Context MCP server (zilliztech/claude-context). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Deterministic rules across all 7 Claude Context tools. Per-identity grants. Full audit log. Live in minutes. Nothing to install.
Free to start. No card required.
7 Claude Context tools catalogued and risk-classified — across an index of 42,500+ MCP servers.