agentic-kv-cache
Reproducing agentic KV-cache policy claims on real traces. 68k requests from 393 Claude Code sessions. LRU is harder to beat than the papers suggest.
Details
- External ID
- 1363812952
- Source
- GITHUB
- Company
- —
- Product
- agentic-kv-cache
- Website domain
- github.com
- Launched
- Sept. 10, 2026
- Cohort
- —
- Upvotes
- 22
- Upvotes percentile
- 0.6369459390212657
- Tags
- kv-cache, llm-agents, llm-inference, prefix-caching, reproducibility, sglang, vllm
- Fetched at
- Sept. 14, 2026, 5:28 p.m.
- Updated at
- Sept. 14, 2026, 5:28 p.m.
Enrichment
- Theme
- low-level systems and developer tools
- Vertical
- Horizontal
- Function
- Observability & eval
- Audience
- Developer
- AI stance
- AI-native
- Project type
- Hobby / open-source project
- Normalized one-liner
- kv-cache benchmark for llm agents
- Manually corrected
- False
Could you build this?
No This is an advanced systems research benchmark measuring KV-cache eviction policies using extensive real-world execution traces across complex LLM context windows.
What it would actually take: Requires building an accurate discrete-event KV-cache memory simulator replicating vLLM/PagedAttention internals, prefix caching trees, and hardware memory tiering. It requires specialized expertise in LLM serving memory hierarchies and access to large proprietary token/session traces from real production agents to benchmark eviction policies.
Competitors
Other products that read as similar to this one — 608 launches clear the similarity bar, closest 8 shown.
Attention rank: #178 of 609 (itself plus its competitors, highest first — normalized so YC and Product Hunt are compared fairly).
Launched 314 days after the earliest competitor.
- UL-SMF · hn · 2026-08-17 · 13 upvotes · similarity 0.50
- OneTriangle - The fastest, cheapest inference, powered by KV cache transfer · yc · 2026-08-21 · 10 upvotes · similarity 0.49
- Cachekit · hn · 2026-01-14 · 47 upvotes · similarity 0.48
- KVBoost · hn · 2026-05-22 · 20 upvotes · similarity 0.48
- Oodle.ai · hn · 2026-07-14 · 31 upvotes · similarity 0.48
- Polygraph · hn · 2026-06-25 · 10 upvotes · similarity 0.47
- Taliesin · hn · 2026-06-04 · 10 upvotes · similarity 0.46
- genpark-agentic-cache-semantic-deduplicator-skill · github · 2026-09-14 · 8 upvotes · similarity 0.45
Other launches for this product
- No other launches for this product.
Same idea, different domain
Nobody's really built a observability & eval tool for Media & entertainment yet.