Early, configurable context auto-compaction for Pi between model responses.

Pre-turn auto-compaction for pi

Session guardian: simulate which Claude Code skills survive auto-compaction

Official auto compaction plugin for dim-agent-sdk.

Resilient context recovery for pi with overflow-safe agentic compaction and guarded continuation after auto-compaction or output-length stops.

A pi extension that caps effective model context windows at 200k tokens for earlier auto-compaction.

A polished input surface for the Pi coding agent with composable statuses, context progress, and highlighted auto-compaction.

Proactive auto-compaction: when the context window crosses a threshold, compact between turns so a long session never hits the wall or needs a restart

RAM-first document database backed by Mataroa and Echo Entries with PocketBase parity. Zero-dependency, append-only, atomic transactions, auto-compaction.

Bridge Claude Code to OpenAI and OpenCode Go models — API keys or ChatGPT/Codex-plan OAuth, with prompt caching, auto-compaction, model switching, and a Claude Code binary patcher

Per-model context-usage thresholds for Pi's built-in auto-compaction, so models with different context windows and performance profiles compact at the right time

Configure Pi auto-compaction using a context-window percentage threshold.

Auto-compaction, auto-continue, auto-planning. Replaces OpenCode's default auto-compact/auto-continue lifecycle with one that creates a portable, persistent PLAN.md document. One persistent source of truth for your project that grows as the agent works au

A TypeScript port of the codex-rs agent core — runs the Codex sampling loop (Responses API streaming + tool calls + auto-compaction) directly in Node or the browser, with no spawned process.

Pi extension: live context-usage bar + debounced auto-compaction at a configurable threshold.

Pi extension that waits for overflow auto-compaction retry before headless prompt completion.

Makes pi's auto-compaction configurable: cap context windows, set compaction trigger targets, improve summarization quality, and manage settings via an interactive /compaction command.

Cap model context windows so pi's built-in auto-compaction fires earlier. Zero-config 200k default for long-context Claude models; configurable for anything else.

Provider-agnostic LLM context manager with tool execution, auto-compaction, reactive state, and an event bus.

pi coding-agent extension: proactive auto-compaction with a configurable, cheap summarization model (formerly deepseek-compact)