AI agents use upload_file to create or update resources in Redmine — usually the action step of a workflow, after the agent has gathered context. Every call changes real data in your Redmine environment.
This tool creates a new file resource and attaches it to a project, modifying the project's state. This is a Write operation because it adds data reversibly—the uploaded file can be removed. It is not Read (no retrieval), not Execute (no code execution or external triggers), not Destructive (not permanent deletion), and not Financial.
From the tool's definition upload_file - 'Upload a file and attach it to a project's Files area'. The verb 'Upload' and 'attach' indicate creation and modification of project data (adding a new file resource). The description explicitly states attachment to a project area.
Attacks that exploit this kind of access
Upload a file and attach it to a project's Files area in one. It is categorised as a Write tool in the Redmine MCP Server, which means it can create or modify data. Consider rate limits to prevent runaway writes.
Register the Redmine MCP server in PolicyLayer and add a rule for upload_file: 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 Redmine. Nothing to install.
upload_file 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 upload_file 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 upload_file. 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.
upload_file is provided by the Redmine MCP server (KalvadTech/redmine-mcp). 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 →