ask-mcp-iteration-limit-is-five
IN premise
When `mcp_servers` is non-empty, `ask()` allows up to 5 tool-call iterations before forcing a final LLM response (6 total invocations), compared to the lower limit without MCP servers.
Summary
When MCP servers are configured, the ask function gives the LLM more room to work — up to 5 rounds of tool calls plus a final response, for 6 total LLM invocations. Without MCP servers, the iteration cap is lower. This matters because MCP tools often require multi-step reasoning chains, so the higher limit prevents the system from cutting off the LLM before it finishes using the tools it has available.
Dependents
These beliefs depend on this one:
- ask-mcp-integration-is-safely-bounded — MCP tool calls in `ask()` are both error-tolerant (exceptions caught and fed back as context for alternative tool selection) and iteration-bounded (5 tool-call rounds max), preventing both crashes from MCP server failures and runaway tool loops.
- ask-mcp-tool-use-has-current-catalog — Ask's MCP tool integration achieves full reliability — errors caught, iterations bounded — with the tool catalog always reflecting the MCP server's current capabilities rather than a stale connection-time snapshot.
Details
| Source | entries/2026/05/11/tests-test_ask_mcp.md |