How the agent loop works changedagent-sdk/agent-loop
Nearest release: v2.1.304, published under an hour before upstream edited the page. Shown because the two are within 24 hours of each other. Nothing here says the release caused the edit.
Upstream edited this page at 19 Oct 2026 19:52 UTC, give or take a minute or two: the time comes from Anthropic’s own sitemap rather than from a commit. This site recorded the change at 19 Oct 2026 20:07 UTC.
Upstream edited
Recorded here
Lines+2added
Lines−2removed
From line
255
where the diff opens
First seen
14 Aug 2026
this site's first read of the page
Recorded edits19to this page, all time
The whole hunk
from line 255, old and new numbered
/
from line 255
255255| :- | :- | :- |
256256| **System prompt** | Every request | Small fixed cost, always present |
257257| **CLAUDE.md files** | Session start, via [`settingSources`](/docs/en/agent-sdk/claude-code-features) | Full content in every request (but prompt-cached, so only the first request pays full cost) |
258| **Tool definitions** | Every request; MCP schemas deferred by default | Built-in tool schemas load every request. [Tool search](/docs/en/agent-sdk/mcp#mcp-tool-search) defers MCP tool schemas by default, falling back to upfront loading on unsupported models and certain platforms. See [Configure tool search](/docs/en/agent-sdk/tool-search#configure-tool-search) for the full matrix |
258| **Tool definitions** | Every request for core built-in tools such as Bash, Read, and Edit; on demand for MCP tools and other built-in tools, which [tool search](/docs/en/agent-sdk/mcp#mcp-tool-search) defers by default | Every schema loads upfront in the cases under [Configure tool search](/docs/en/agent-sdk/tool-search#configure-tool-search) |
259259| **Conversation history** | Accumulates over turns | Grows with each turn: prompts, responses, tool inputs, tool outputs |
260260| **Skill descriptions** | Session start, via setting sources | Short summaries; full content loads only when invoked |
261261
from line 291
291291
292292A few strategies for long-running agents:
293293
294* **Use subagents for subtasks.** Each subagent starts with a fresh conversation (no prior message history, though it does load its own system prompt and project-level context like CLAUDE.md). It does not see the parent's turns, and only its final response returns to the parent. The main agent's context grows by that summary, not by the full subtask transcript. See [What subagents inherit](/docs/en/agent-sdk/subagents#what-subagents-inherit) for details.
294* **Use subagents for subtasks.** A subagent returns only its final response, so your main agent's context grows by a summary, not by the subtask's transcript.
295295* **Be selective with tools.** Every tool definition takes context space. Use the `tools` field on [`AgentDefinition`](/docs/en/agent-sdk/subagents#agentdefinition-configuration) to scope subagents to the minimum set they need.
296296* **Watch MCP server costs.** [MCP tool search](/docs/en/agent-sdk/mcp#mcp-tool-search) defers MCP tool schemas by default and loads them on demand. When tool search is off or has fallen back to upfront loading, each MCP server adds all its tool schemas to every request, so a few servers with many tools can consume significant context before the agent does any work. See [Configure tool search](/docs/en/agent-sdk/tool-search#configure-tool-search) for the configurations where the fallback applies.
297297* **Use lower effort for routine tasks.** Set [effort](#effort-level) to `"low"` for agents that only need to read files or list directories. This reduces token usage and cost.
No line in this hunk matches that.