Untested on omp

Configurable, cheap smart context management for the Pi coding agent

从知识库文档（PDF、报告、日报、Markdown 等）里精准检索相关段落：融合关键词（词法） 与向量（语义）两路召回，再用 MMR 去重重排。当用户要在知识库中查找资料、检索或搜索 文档内容时使用，触发词：检索、知识库、文档、搜索、查找、查资料、找资料、找文档、 搜文档、知识库检索、检索知识库、混合检索、向量检索、语义检索、RAG。也用于「搜一下知识库」 「查一下XX文档」「知识库里有没有关于XX的内容」「帮我找XX相关的文档」等短语。 不用于：通用网页搜索（WebSearch）、飞

Freeze default model/thinking in settings.json so switching models or toggling thinking never rewrites the defaults.

Pi extension: load a Simocracy sim into your chat — see its pixel-art sprite render inline in the terminal and roleplay with it.

An adaptive capability router for Pi with progressive tool-guideline and Skill distillation.

Pi extension scaffolding for showing subscription and quota usage across AI providers.

Prompts for confirmation before running dangerous commands in the pi-coding-agent.

Arcade minigames for the Pi coding agent.

Intent-aware next-prompt suggestion extension for pi

A pi extension that intercepts potentially dangerous bash commands and prompts for confirmation before execution

Public Pi-Rogue package containing the explicit Advisor and read-only Board extension.

Give Pi a lightweight sense of time with low-noise anchors and exact time on demand.

Search and query past pi sessions. Recall previous conversations on demand.

Compact, high-signal rendering for Pi tool calls and results.

Pi extension package for browser automation with Puppeteer-core, across the Chromium family and Firefox.

Pi coding agent input translation extension using Bergamot language pair models.

Live UniPi trajectory inspector for Pi sessions

Generate and edit images using Codex CLI

TypeSafe ranking, classification, retrieval and structured-decision tools for Pi coding agents

Delegate pi work to isolated, budgeted children with verification loops and tracing.

Declarative flag-presence and flag-allowlist predicates for pi-steering rules. First official external plugin.

Shorten long URLs and file paths in Pi assistant output so OSC 8 / click targets fit the pane width.

Shared usage data core for pi extensions

Pi extension that shares sessions to a self-hosted Opengist instance

Workspace & context-management skills — repo catch-up, AGENTS.md initialization, ecosystem context-file architecture, doc organization, planning sync, and cross-session handoff. Host-neutral: any AGENTS.md-aware coding agent.

Pi extension for Kimi K2 Coding plan — OAuth-based provider with kimi-for-coding model

Powerline statusline footer for the pi coding agent — built-in themes, spacious full + multiline layouts, rich git/PR/context segments.

Pi extension providing access to Cursor models via OAuth and a local OpenAI-compatible gRPC proxy

Enhanced Pi extension for Charm Hyper with dashboard, live rate limits, and usage analytics

Automated code review loop extension for pi coding agent

Deterministic tool-call guardrails for pi — enforce rules with before-tool hooks instead of prompts. Zero token cost, 100% reliability.

Lazy senior dev mode for AI agents. The best code is the code you never wrote.

Herdr-backed subagents for Pi: visible, steerable, resumable child agents in Herdr panes

Graft context-graph support for the pi coding harness — Claude-parity hooks, CLI-backed tools, skill, statusline and commands

Live jj (jujutsu) working copy status and repo context for pi

Pi-native extension: gradient descent for your AGENTS.md — evidence-backed, human-gated (inspired by backpass)

Copy the current Pi prompt editor text to the system clipboard

Pi coding agent provider extension exposing an Anthropic-compatible proxy (via ANTHROPIC_BASE_URL / ANTHROPIC_API_KEY) with Claude Sonnet 5, Opus 5, and Haiku 4.5 models.

Floating Editor, summon your favourite agents for help, all in zellij

Web UI for the pi coding agent (OpenWebUI-style, mobile-friendly)

A cohesive global visual system for the Pi terminal UI

Multi-agent orchestration suite for Pi — 6 modes, 43 extensions, 11 themes

Evidence-backed review of AI-assisted TypeScript, JavaScript, and Python changes for Pi

Pi extension — lazy-launch Chrome with Playwright, persistent profile, stealth patches, simple browser tools

Pi agent extension for Grafana Agent Observability

Ghost-text prompt suggestions (inline autocomplete + next-prompt) for the pi coding agent

Unconditionally blocks destructive bash commands before execution. A pi extension that guards against rm, git reset --hard, docker rm, aws delete operations, and more.

Agents should invoke this skill whenever they write or edit code in permanent files, regardless of language or change size. Improve code directly within the authorized task while preserving explicit review-only requests.

Community Pi integration for the official MemPalace core.

Production-grade academic research pipeline for pi: research → write → review → revise → finalize. 4 skills, 27 modes, 39-agent ensemble.