Retry a failed fine-tuning job from its last checkpoint. Creates a new job that resumes training where the failed one stopped, saving GPU time. Each retry is billed separately.\n\n
AI agents call retry_job to retrieve information from Tuning Engines - LLM Fine-Tuning without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
Even though retry_job only reads data, uncontrolled read access leaks sensitive information and racks up API costs — an agent caught in a retry loop can make thousands of calls a minute without anyone noticing.
Attacks that exploit this kind of access
Retry a failed fine-tuning job from its last checkpoint. Creates a new job that resumes training where the failed one stopped, saving GPU time. Each retry is billed separately.\n\n. It is categorised as a Read tool in the Tuning Engines - LLM Fine-Tuning MCP Server, which means it retrieves data without modifying state.
Register the Tuning Engines - LLM Fine-Tuning MCP server in PolicyLayer and add a rule for retry_job: 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 Tuning Engines - LLM Fine-Tuning. Nothing to install.
retry_job is a Read tool with low risk. Read-only tools are generally safe to allow by default.
Yes. Add a rate_limit block to the retry_job 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 retry_job. 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.
retry_job is provided by the Tuning Engines - LLM Fine-Tuning MCP server (tuningengines-cli). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.