Push multiple files to a GitHub repository in a single commit
AI agents use push_files to create or update resources in MCP-Brave-Search — usually the action step of a workflow, after the agent has gathered context. Every call changes real data in your MCP-Brave-Search environment.
The tool modifies repository content by pushing files, which is a Write operation. Severity is high because unauthorized pushes could inject malicious code, alter source code, or compromise a repository's integrity, affecting all consumers of that repository. However, it is not Destructive since pushes are reversible (can be reverted) and not Financial. Confidence is high given the clear description.
From the tool's definition Tool description states 'Push multiple files to a GitHub repository in a single commit' — this creates or modifies data in a repository, a reversible action.
Documented attack patterns abuse exactly the kind of access push_files gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and MCP-Brave-Search, and nothing reaches the server without passing your rules. This is the rule we recommend for push_files:
{
"version": "1",
"default": "deny",
"tools": {
"push_files": {
"limits": [
{
"counter": "push_files_rate",
"window": "minute",
"max": 30,
"scope": "grant"
}
]
}
}
}push_files stays usable, but capped — an agent stuck in a loop can't make hundreds of changes a minute. Everything else on the server is denied unless you say otherwise.
Free to start. No card required.
Push multiple files to a GitHub repository in a single commit. It is categorised as a Write tool in the MCP-Brave-Search MCP Server, which means it can create or modify data. Consider rate limits to prevent runaway writes.
Register the MCP-Brave-Search MCP server in PolicyLayer and add a rule for push_files: 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-Brave-Search. Nothing to install.
push_files 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 push_files 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 push_files. 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.
push_files is provided by the MCP-Brave-Search MCP server (modelcontextprotocol/servers-archived). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Start from MCP-Brave-Search, add the rest of your stack, and see everything your agents can call. Then put policy on all of it.
Free to start. No card required.
59 MCP-Brave-Search tools catalogued and risk-classified — across an index of 43,000+ MCP servers.