bug(mcp): orchestration_wait_agents reply is lost when the caller backgrounds a long wait #827

Expected behavior

A caller of orchestration_wait_agents with a long timeout receives the result when the children finish, or a bounded "still running" reply it can poll on.

Actual behavior

On 2026-09-05T23:28Z a Claude session waited on a two-child run with a generous timeout. The harness moved the MCP call to the background at 120 s; the MCP transport then dropped mid-call and the tool's response was lost (MCP server "agent_code" transport dropped mid-call; response for tool "orchestration_wait_agents" was lost). The children were unaffected. A retry with a 100 s timeout returned normally.

Impact

Any orchestration flow that waits longer than the harness's backgrounding threshold silently loses its wait result and must fall back to polling orchestration_list_agents. Developer-experience reliability failure in a first-party MCP surface.

Proposed fix

Cap the server-side wait below the harness threshold (90 s) regardless of the requested timeout and return { timedOut: true, agents: [...] } so callers poll; document the cap in the tool description.

Environment

Agent Code 0.0.2-beta.1 (main at 5d64184 at the time), Claude Code harness with a 120 s foreground tool limit.