Add HTTP size limits

docs/operations.md

@@ -36,6 +36,9 @@ Built-in defaults relevant to operations:
 - `schema_dir: .`
 - `server.addr: :8080`
 - `server.artifact_root`: unset; HTTP `file` input references are rejected until configured
+- `server.max_request_bytes: 16777216`
+- `server.max_artifact_bytes: 16777216`
+- `server.max_response_bytes: 16777216`
 - `defaults.render_format: text`
 
 ## Normal CLI Workflow
@@ -78,6 +81,7 @@ Current inbound API behavior:
 - Validation content failures still return `200 OK` with `validation.status: "failed"`.
 - `inline` input references work without filesystem configuration.
 - `file` input references require `server.artifact_root` or `serve --artifact-root`; relative paths resolve inside that root and paths outside it are rejected.
+- Request bodies, HTTP file input artifacts, and encoded JSON responses are limited by `server.max_request_bytes`, `server.max_artifact_bytes`, and `server.max_response_bytes`.
 
 Security caveat:
 
@@ -85,6 +89,13 @@ Security caveat:
 - Deploy only behind trusted controls (private network boundary, authenticated reverse proxy, API gateway, or equivalent).
 - Keep the HTTP artifact root as narrow as practical and do not make it writable by untrusted users.
 
+Sizing guidance:
+
+- Keep limits at the defaults unless a deployment has a measured need for larger prompt inputs or outputs.
+- Prefer `inline` inputs for small payloads and HTTP `file` inputs for larger local artifacts inside a controlled artifact root.
+- Increase the response limit when prompts intentionally return large generated artifacts or when clients request `include_raw_output`.
+- Set a limit to `0` only for trusted deployments where another layer enforces request and response size.
+
 ## Output, Logs, And Exit Codes
 
 `run` command: