AI agents invoke eval_code to trigger actions in Pharo Smalltalk Interop. 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.
'eval_code' almost certainly evaluates arbitrary Smalltalk code in the local Pharo image, which constitutes arbitrary code execution. The blast radius is critical as it could read/write/delete files, modify the running image, or execute system commands. Confidence is slightly reduced due to the empty description, but the name and server context make execution semantics highly probable.
From the tool's definition Tool name 'eval_code' on a Pharo Smalltalk interop server strongly implies arbitrary code evaluation in the Smalltalk image; sibling tool 'eval' confirms code execution is a primary capability of this server.
Documented attack patterns abuse exactly the kind of access eval_code gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and Pharo Smalltalk Interop, and nothing reaches the server without passing your rules. This is the rule we recommend for eval_code:
{
"version": "1",
"default": "deny",
"tools": {
"eval_code": {
"limits": [
{
"counter": "eval_code_rate",
"window": "minute",
"max": 10,
"scope": "grant"
}
]
}
}
}eval_code 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.
eval_code. It is categorised as a Execute tool in the Pharo Smalltalk Interop MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the Pharo Smalltalk Interop MCP server in PolicyLayer and add a rule for eval_code: 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 Pharo Smalltalk Interop. Nothing to install.
eval_code 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 eval_code 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 eval_code. 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.
eval_code is provided by the Pharo Smalltalk Interop MCP server (mumez/pharo-smalltalk-interop-mcp-server). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Start from Pharo Smalltalk Interop, 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.
23 Pharo Smalltalk Interop tools catalogued and risk-classified — across an index of 43,000+ MCP servers.