pi-prefix-cache-compaction
MITPi compaction that reuses the model server's prefix cache (vLLM, SGLang, llama.cpp, or hosted APIs with automatic caching such as DeepSeek; Anthropic Messages and OpenAI Chat Completions): no cold re-prefill of the history, plus a warm-up so the next turn
100
Security score
0 known advisories in v0.3.0
Weekly downloads
291
Unpacked size
54.1 kB
Dependencies
0
Last publish
10 hours ago
Security advisories
No known vulnerabilities affect v0.3.0.
Downloads — last 30 days
291 total
Aug 22Sep 20