fast-jev-compaction alternatives
Curated alternatives to fast-jev-compaction — and why you'd switch.
token-optimizer
Context-waste hunter for coding agents: hooks compress reads, bash and search output, checkpoint before compaction, and audit waste in configs, skills, MCP and memory — with a local dashboard.
Why switchBoth fight context waste in Claude Code with hooks; token-optimizer compresses reads, bash and search output on the way in, fast-jev-compaction leaves everything verbatim and instead deletes whole tool calls a decision model judges stale.
Full comparison →headroom
Context compression layer for agents: squeezes tool outputs, logs, files and RAG chunks 20-95% before the LLM — reversible, local-first; library, proxy, one-command agent wrap, or MCP.
Why switchSame goal, opposite method: headroom squeezes tool output, logs and chunks 20-95% before the model sees them, this drops entire calls and keeps survivors byte-for-byte - lossy-but-smaller versus exact-but-fewer.
Full comparison →SoL-Pi
NVIDIA's Pi extension with four auto-research-discovered efficiency mechanisms - action fusion, observation packing, evidence-preserving log reduction, online context compaction - all opt-in.
Why switchsol-pi's ObservationPack and evidence-preserving reducer solve the same replay problem inside Pi with quotation checks; fast-jev-compaction solves it in Claude Code by scoring each call with Jev.
Full comparison →