[Automated] Draft docs (agentgateway): llm: check if format supports guards before attempting response guard

[github-actions]

agentgateway/agentgateway#3555 — llm: check if format supports guards before attempting response guard

Docs versions: standalone/main, kubernetes/main

What changes for users: Response guards are skipped when the request input format does not support prompt guards. The request side already had this support check; the response side now follows the same boundary.

What the docs now say: The standalone prompt guards overview and Kubernetes guardrails overview now state that response guards run only when the request input format supports prompt guards.

Not verified: No cluster; the steps are unrun.

How agentgateway-3555 was drafted, and what was not verified

• Code PR: llm: check if format supports guards before attempting response guard agentgateway#3555
• Code issue: Requests sent to Detect route types with promptGuard configured causes connection close / panic agentgateway#3552
• Product: agentgateway (upstream) → agentgateway/website
• Docs version: standalone/main and kubernetes/main
• Category: scoped-edit / pr

Plan, and what changed it

• Planned: Add the response-guard input-format boundary to both guardrails overview pages that the code-path rule named.
• Learned: The neighbor dossiers do not carry the same behavior: #3530 is guardrail rejection logging, #3554 is build flags, and #3558 is streaming copy behavior.
• Did: Added the boundary sentence to both overview pages and wrote a proposed release note.

Why a documentation change is necessary

The pull request adds req.input_format.supports_prompt_guard() before applying a response prompt guard. The PR body says the check existed only on the request side, while both target pages said response guards could reject or mask responses without naming the input-format boundary. A reader configuring response guards could expect them to run on traffic whose format cannot support prompt guards.

What changed on disk

• content/docs/standalone/main/documentation/llm/prompt-guards/overview.md — Added the response-guard input-format boundary.
• content/docs/kubernetes/main/documentation/llm/guardrails/overview.md — Added the response-guard input-format boundary.

How to verify this change

1. Configure a response guard on an LLM route.
2. Send an LLM request in a format that does not support prompt guards.
3. Confirm that the response guard does not run, then repeat with a supported request input format and confirm that the response guard runs.

What was not verified

The configuration was not applied to a cluster, so every command in this diff is unrun. The dossier does not name a concrete input format that returns supports_prompt_guard() == false, so the exact traffic shape for the unsupported-format check was not verified.

How the pages were chosen (triage report)

Draft a documentation change

agentgateway/agentgateway#3555 — llm: check if format supports guards before attempting response guard

• Product: agentgateway (upstream) → agentgateway/website
• Docs version: standalone/main, kubernetes/main
• Code issue: Requests sent to Detect route types with promptGuard configured causes connection close / panic agentgateway#3552
• Summary (read from the body): This was only on the req side
• Not tested: no cluster was available, so every command in this diff is unrun

Warnings

• llm: check if format supports guards before attempting response guard agentgateway#3555: re-drafting because it was named on this run, though its earlier draft is issue. Nothing removes the earlier entry; the new draft replaces it in the ledger when it opens.

Draft branch: pr-tracker-draft-agentgateway-3555

[cloudflare-workers-and-pages]

Deploying agentproxy with    Cloudflare Pages

Latest commit:	04a0015
Status:	✅  Deploy successful!
Preview URL:	https://be68aa71.agentproxy.pages.dev
Branch Preview URL:	https://pr-tracker-draft-agentgatewa-4ksx.agentproxy.pages.dev

View logs