Bazel build system operations: build, test, query, info, run, clean, fetch.
AI agents invoke bazel to trigger actions in Npm. 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.
The 'run' and 'build' operations are direct code execution: they invoke the Bazel build system to compile and execute targets. The 'test' operation runs arbitrary test suites. These are not simple read operations (like 'query' or 'info') but active executions whose outcomes depend on user-supplied arguments and the underlying build configuration.
From the tool's definition Tool provides 'run', 'build', 'test', and 'fetch' operations via the Bazel build system. These are code execution primitives that can trigger arbitrary build processes, execute test suites, and fetch external dependencies, whose effects depend entirely on the…
Documented attack patterns abuse exactly the kind of access bazel gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and Npm, and nothing reaches the server without passing your rules. This is the rule we recommend for bazel:
{
"version": "1",
"default": "deny",
"tools": {
"bazel": {
"limits": [
{
"counter": "bazel_rate",
"window": "minute",
"max": 10,
"scope": "grant"
}
]
}
}
}bazel stays usable, but rate-capped — a runaway agent can't fire it dozens of times a minute. Everything else on the server is denied unless you say otherwise.
Free to start. No card required.
Bazel build system operations: build, test, query, info, run, clean, fetch. It is categorised as a Execute tool in the Npm MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the Npm MCP server in PolicyLayer and add a rule for bazel: 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 Npm. Nothing to install.
bazel 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 bazel 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 bazel. 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.
bazel is provided by the Npm MCP server (Dave-London/Pare). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Start from Npm, 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.
202 Npm tools catalogued and risk-classified — across an index of 43,000+ MCP servers.