AI agents use accept_differential to create or update resources in Phabricator MCP Server — usually the action step of a workflow, after the agent has gathered context. Every call changes real data in your Phabricator MCP Server environment.
Accepting a differential revision modifies its state in the code review workflow (marking it as approved/accepted), which is a reversible write action. However, it carries high severity because accepting code reviews can directly lead to code being merged, with significant downstream consequences if misused by an AI agent.
From the tool's definition Accept a differential revision
Documented attack patterns abuse exactly the kind of access accept_differential gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and Phabricator MCP Server, and nothing reaches the server without passing your rules. This is the rule we recommend for accept_differential:
{
"version": "1",
"default": "deny",
"tools": {
"accept_differential": {
"limits": [
{
"counter": "accept_differential_rate",
"window": "minute",
"max": 30,
"scope": "grant"
}
]
}
}
} accept_differential 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.
Accept a differential revision. It is categorised as a Write tool in the Phabricator MCP Server MCP Server, which means it can create or modify data. Consider rate limits to prevent runaway writes.
Register the Phabricator MCP Server MCP server in PolicyLayer and add a rule for accept_differential: 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 Phabricator MCP Server. Nothing to install.
accept_differential 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 accept_differential 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 accept_differential. 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.
accept_differential is provided by the Phabricator MCP Server MCP server (yushengauggie/phabricator-mcp-server). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Start from Phabricator MCP Server, 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.
11 Phabricator MCP Server tools catalogued and risk-classified — across an index of 43,000+ MCP servers.