docs(headroom): clarify that single-turn requests are never compressed

[devin-ai-integration]

Summary

The Headroom quick start told people to test with a system message plus one user message, which can never compress anything. LiteLLM holds back the system rows, the last user row, and the last assistant row (get_protected_indices in litellm/compression/compress.py, applied by the Headroom guardrail's _protected_indices), so on that payload compressible is empty and the guardrail returns before ever calling /v1/compress. Anyone following the page verbatim sees Headroom do nothing and reasonably concludes the integration is broken.

Reproduced on a live proxy (guardrails: ["headroom-compression"], also attached to a virtual key): the doc's own curl returns 200 with zero requests reaching the Headroom service, while the same call with one earlier user/assistant turn sends 3 rows to /v1/compress.

Changes: the quick-start examples (chat completions and /v1/messages) now carry an earlier turn, a paragraph after them explains the protected rows and the single-turn no-op, and the "Why requests_compressed can be 0" section now starts with that check before the Headroom-container defaults.

Also corrected two claims about x-litellm-applied-guardrails. That header is written from the guardrails the request opted into (_process_guardrail_metadata in litellm/proxy/utils.py), so it is present even when Headroom was skipped; it confirms opt-in, not that any message was rewritten. guardrail_information on the spend log row is the signal for the latter.

Link to Devin session: https://app.devin.ai/sessions/835118e39ed94d4cbeafc573bfd2d078

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment, CI, and merge conflict monitoring

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
litellm	Ready	Preview	Aug 24, 2026 11:01pm