AI agents invoke git_checkout to trigger actions in Mcp Git. What it does depends on the arguments the agent supplies, and its effects often reach beyond the immediate call — builds kicked off, notifications sent, workflows started.
git_checkout switches the active branch or restores working tree files. While not purely destructive (branches still exist), it triggers external filesystem operations that can overwrite local changes and alter repository state. It spans Write/Execute; Execute is the most appropriate as it triggers a repo operation with side effects beyond simple data modification.
From the tool's definition 'Switches branches' — changing the working tree state is an external operation that modifies the working directory and HEAD pointer, potentially overwriting uncommitted changes
Attacks that exploit this kind of access
Switches branches. It is categorised as a Execute tool in the Mcp Git MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the Mcp Git MCP server in PolicyLayer and add a rule for git_checkout: 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 Git. Nothing to install.
git_checkout is a Execute tool with high risk. Execute tools should be rate-limited and have argument validation enabled.
Yes. Add a rate_limit block to the git_checkout 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 git_checkout. 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.
git_checkout is provided by the Mcp Git MCP server (mcp-git). 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 →