Context Pager

Cut your AI agent's token costs by 4-10x.

Context Pager lets your AI read compressed pages of documents instead of full files. Like an OS pages memory to RAM, your agent pages through documents on demand.

Three tools

• search_documents — find relevant documents and recalled insights
• compress_document — read a compressed, PII-masked page
• commit_to_long_term_memory — persist facts that resurface on later searches

Get started

	Local only	With relay	Self-hosted
Time	2 min	5 min	15 min
Internet	Not needed	First time	Required
Best for	Trying it out	Multiple devices	Teams

pip install "context-pager[bridge]"

Sign up

curl -X POST https://context-pager.duckdns.org/v1/signup

Returns your pgr_agent_* and pgr_bridge_* keys. Shown once.

---

GitHub · Full setup guide · Architecture & API