Lean output compaction for terminal-heavy agent workflows.

Helpers for context windows, usage normalization, compaction, and cost estimation.

Umbrella package: installs all first-party Prism compaction strategies. Imports come from the individual @arnilo/prism-compaction-* packages.

LLM-backed compaction strategy package for Prism.

Observational memory compaction package for Prism.

No description provided.

Model-summarize-then-replace conversation compaction

Pi extension for goosedump session search, durable memory, and compaction

Async context compaction for the Pi coding agent: precompute background summaries for long sessions.

OpenClaw tokenjuice exec output compaction plugin

Pi extension that optimizes RTK command rewriting and tool output compaction for the coding agent.

Pi extension for proactive model-aware context compaction thresholds

Choose a dedicated model for Pi's native compaction without changing its compaction algorithm or prompts.

Pillar-2 active context curation — pre-compaction relevance + dedupe pass that lets compaction-autocompact skip the model call when curation alone brings tokens under the trigger threshold

Advanced compaction extension and skill for Pi with automatic threshold-based compaction and critical context preservation

Layered, local, vector-backed context compressor for pi — supersede/collapse/cluster compaction with deduped inline recall.

Better compaction for pi: native /responses/compact replay for OpenAI Responses APIs, plus a configurable compaction model driving pi's native summarization everywhere else.

Dedicated Pi package for custom session compaction summaries.

Observational memory extension for pi — cache-friendly tiered compaction with observations and reflections.

Context-window compaction plugin for the agents runtime.