AI agents use apply_patch to create or update resources in Mcp Flow — usually the action step of a workflow, after the agent has gathered context. Every call changes real data in your Mcp Flow environment.
apply_patch modifies project files by applying diffs, making it a Write operation (creates or modifies data reversibly). It is not Execute because it doesn't run code or external operations—it only applies pre-computed diffs. It is not Destructive because diffs are typically reversible and the default dry-run mode prevents accidental application.
From the tool's definition Tool name 'apply_patch' and description 'Apply a unified diff to the project' indicates modification of project files. Default dryRun=true suggests reversibility by design, but actual application modifies files.
Attacks that exploit this kind of access
Apply a unified diff to the project. Defaults to dryRun=true. All. It is categorised as a Write tool in the Mcp Flow MCP Server, which means it can create or modify data. Consider rate limits to prevent runaway writes.
Register the Mcp Flow MCP server in PolicyLayer and add a rule for apply_patch: 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 Mcp Flow. Nothing to install.
apply_patch is a Write tool with medium risk. Write tools should be rate-limited to prevent accidental bulk modifications.
Yes. Add a rate_limit block to the apply_patch 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 apply_patch. 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.
apply_patch is provided by the Mcp Flow MCP server (remimenguy/mcp-flow). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Every MCP server has a record like this.
Type a name, get the same breakdown: verified identity, auth posture, risk grade, capabilities, recommended policy.
Teams ship this data inside their own products. See what a licence covers →