{"version":"2.1.302","anchor":"mcp-tool-results-report-shell-cpu-from-the-cgroup","canonical_anchor":"mcp-tool-results-report-shell-cpu-from-the-cgroup","heading":"MCP tool results can now report how much CPU the shell used","tier":"internal","area":"MCP","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.302\/e\/mcp-tool-results-report-shell-cpu-from-the-cgroup","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.302","markdown":"### MCP tool results can now report how much CPU the shell used\n\nMCP tool responses can now include a shell CPU time figure, read from the system's cgroup accounting when the value looks sane\n\n**What**\n\nMCP servers are outside programs that give Claude Code extra tools. Claude Code already attached timing details to MCP tool responses, such as start time, uptime and CPU time.\n\nIt now also reads the file `cpuacct.usage` from the cgroup. A cgroup is a Linux feature that groups processes and tracks the resources they use. When the reading looks sane, Claude Code adds `shell_cpu_ms`, the CPU time used by the shell in milliseconds, to those details.\n\n**Why**\n\nHosted environments get a measure of how much CPU the shell used, next to the timing figures that were already reported.\n\n- Area: MCP\n- Tier: Under the hood\n- Useful: 1\/5\n- Signal: 2\/5\n- Scope: individual\n- Heads-up: no"}