> ## Documentation Index
> Fetch the complete documentation index at: https://gateway.consus.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Agent Harnesses & Context Windows

> Sizing agent loops against the gateway's real context ceilings, and configuring Claude Code, the Claude Agent SDK, and custom loops to stay under them

Agent harnesses — Claude Code, the Claude Agent SDK, OpenAI-SDK tool loops — grow a conversation on every turn. Long sessions eventually approach the model's **real context ceiling on the platform serving it**, which is not always the number the model's commercial documentation advertises. A harness that believes the wrong ceiling never compacts proactively; it discovers the truth as a `400` on a live request.

## Effective context windows

The limit that matters is the **input (prompt) budget enforced by the gov cloud platform** behind the route — the `context_window` in the [Model Explorer](/models) is the authoritative per-model figure. The ceilings agent sessions most commonly reach:

| Model                                                                        | Effective input limit                 |
| ---------------------------------------------------------------------------- | ------------------------------------- |
| `claude-sonnet-4-5`, `claude-haiku-4-5`                                      | **200K tokens**                       |
| `claude-opus-4-8`, `claude-opus-4-7`, `claude-opus-4-6`, `claude-sonnet-4-6` | **1M tokens**                         |
| `gpt-5.4`                                                                    | **272K tokens** (enforced at 278,528) |

## What hitting the ceiling looks like

A request over the input limit returns `400 invalid_request_error` with `code: "context_window_exceeded"` — including on `stream: true` requests, where it arrives as a real HTTP 400 before any events. When the provider reports them, the message carries the attempted and maximum token counts. This error is **terminal for the request as sent**: retrying it verbatim can never succeed. Trim, summarize, or compact the conversation and resend. See [Errors](/errors) for the envelope.

## Claude Code and the Claude Agent SDK

Claude Code determines when to auto-compact from what it believes the model's context window is — via a built-in table of **stock Anthropic model ids**. The gateway's composite ids (`claude-sonnet-4-5:il5+itar`, `claude-opus-4-8:fedramp-moderate`, …) are not in that table, so calibrate the client explicitly:

```bash theme={null}
# Match the served model's input limit from the table above
# (example: a claude-sonnet-4-5 route)
export CLAUDE_CODE_AUTO_COMPACT_WINDOW=200000
```

With the window set, proactive auto-compaction fires before the ceiling and sessions stop bouncing off `400`s. Two more rules:

* **Only claim the window the route actually serves.** Appending `[1m]` to a model id makes the client assume a 1M-token window — on a 200K route (e.g. `claude-sonnet-4-5:il5+itar`) that disables proactive compaction entirely, and on the 1M models it adds nothing over setting the window explicitly.
* **Pre-count when precision matters.** `POST /v1/messages/count_tokens` is supported on the Claude routes (both Bedrock- and Vertex-served) and returns the exact input count for a candidate request.

## Custom loops (OpenAI or Anthropic SDK)

If you run your own agent loop, two behaviors keep it healthy against the gateway:

1. **Track context from response usage.** Each response's `usage` block reports actual input tokens; budget the next turn against the table above, leaving headroom for `max_tokens` and any thinking budget.
2. **Classify before retrying.** Retry `429`, `502`, `503`, and timeouts with backoff. Never retry a `400` — and specifically treat `code: "context_window_exceeded"` as a signal to compact, not resend. Streamed requests now surface pre-stream failures with their real HTTP status, so ordinary SDK status handling works; only errors arriving mid-stream (after events have flowed) come as in-band SSE error frames.

## Compaction options

| Approach                                      | Where it runs      | Notes                                                                                                           |
| --------------------------------------------- | ------------------ | --------------------------------------------------------------------------------------------------------------- |
| Claude Code / Agent SDK auto-compact          | Client             | Calibrate with `CLAUDE_CODE_AUTO_COMPACT_WINDOW` as above                                                       |
| `context_management` (clear old tool results) | Server (Anthropic) | Accepted on the Claude gov routes via `/v1/messages`; clears stale tool output before the model sees the prompt |
| Application-level summarization               | Client             | Summarize or drop old turns in your own loop before the ceiling                                                 |