Untested on omp
-
Compact animated reasoning previews for Pi
-
[DEPRECATED] Delegate work to Claude Code — use pi-harness-delegate (pi install npm:pi-harness-delegate) — shim re-exporting claude harness.
-
Pi extension: multi-model coding decisions via OpenRouter
-
Pi extension that starts every session in plan mode: read-only investigation, then a proposal before any file changes.
-
Hide tool rows by default and toggle a compact one-line view.
-
Capability-mode model router for Pi with quality/cost selection and sticky sessions.
-
Search Google and read X (Twitter), Reddit, Amazon, and Google Scholar in your own visible Chrome — same approach as @antirez's ds4-agent. Zero dependencies, no API keys.
-
Durable timing, token, and cost accounting for pi: compact transcript rows after every model request, tool call, compaction, branch summary, and multi-step agent run; a live below-editor activity widget while pi is busy; and /timings for a per-operation i
-
pi coding agent extension: zero-switch image routing for text-only models — transcribes image blocks through a configurable VLM chain before every LLM call (parallel, deduped, cached, with fallback)
-
Pi extension that removes tool output the agent no longer needs from its context. The Jev classifier keeps the needed lines verbatim; stale small outputs, unrelated old exchanges and old screenshots become one-line notes. Everything can be recalled.
-
Pi package scaffold for pi-prompt-form
-
Pi coding agent extension — floating terminal overlay (fish, lazygit, yazi) powered by node-pty
-
UI-free action registry capability for Pi extensions
-
MCP Server for Xiaohongshu (Little Red Book) content creation and publishing automation
-
Auto-discover Aperture models: point pi at an Aperture endpoint and every served model appears in /model, no models.json copy/paste.
-
Quota provider data fetching and caching layer for Pi extensions
-
Lazy senior dev mode for AI agents. The best code is the code you never wrote.
-
Portable Agent Skill for governing lightweight and complex feature-development workflows, including planning, delegated implementation, integration, review, reporting, and completion gates.
-
Curated themes for the Pi coding agent: Rosé Pine (dawn/main), Tokyonight (day/night), Catppuccin (latte/frappé/macchiato/mocha) and Solarized (light/dark) — light & dark variants ready for automatic terminal-following switching.
-
Ollama (local + Ollama Cloud) and LM Studio providers for the pi coding agent - live model discovery with real capabilities (tools/vision/thinking) and context windows.
-
Deprecated: superseded by @blackbelt-technology/pi-dashboard-craft-apps (PhotoCraft plus 12 more ArtCraft Crafting Apps). This shim depends on it and forwards the pi-photocraft bin so existing .pi/mcp.json entries keep working.
-
dsh-style thinking display for pi: one-line header with token count (configurable decimals, default 2); collapsed thinking shows a single-line content preview
-
Pi extension that helps prevent secrets from leaking
-
Hindsight self-hosted memory extension for Pi
-
Claude Code-style TUI output with some personal touches and handy utilities.
-
Currently mid development. Coming soon.
-
Pi package that automatically trusts all project folders.
-
Sync Pi agent config through Git
-
Pi extension for configuring multiple local LLM inference engine connections.
-
Pi extension: local structured mirror of official docs published as llms.txt / llms-full.txt, with differential (ETag/hash) syncing, async TTL refresh, and zero-latency local search/read tools for the agent.
-
A sliding context window for Pi coding agents. When a chat gets long, old messages are saved to a small local database instead of being dropped — so your agent can search them and pull the exact one back whenever it needs it.
-
Compact footer for the pi coding agent — model/thinking and context usage on the left, MCP server count on the right, extension statuses below
-
Test-driven development with red-green-refactor loop. Use when user wants to build features or fix bugs using TDD, mentions "red-green-refactor", wants integration tests, or asks for test-first development.
-
A Pi extension that lets the LLM edit its own context to keep the next prompt lean.
-
Pi package that connects Pi to the Figma desktop MCP server and exposes Figma design tools inside Pi.
-
Pi package scaffold for pi-prompt-guard
-
Persistent Command Code usage line (5h / weekly / monthly quota bars) for the pi coding agent — auto-refreshing widget above the default footer
-
schedule_wakeup / cancel_wakeup for the pi coding agent: session-local self-pacing, the pi analog of Claude Code's ScheduleWakeup dynamic mode.
-
Pi agent-to-agent team communication via shared SQLite message bus with MAMORU guardian — fork of pi-teammate with identity-takeover rejoin fix, presence leases, and an outbound watchdog
-
Reliable scheduled loops for the Pi coding agent — prompt/notify/shell/message on interval, once, or cron. Session-scoped by default (fires only where created), global opt-in. Always-visible status, working input, bulletproof recurrence.
-
thalix-auto — multi-agent orchestration skill for Claude Code, Codex CLI, and Pi (pi.dev). Forked from oh-my-claudecode.
-
Scientific-method R&D orchestration for PI Coding Agent — methodology skills, subagent definitions, /rnd-start pipeline, /rnd-doctor, and a composable tool_call gate registry
-
pi-di18n: full TUI i18n for pi (continues pi-i18n) plus summary localization (integrates pi-compaction-i18n)
-
pi and Oh My Pi (OMP) coding-agent extension that suggests the next logical prompt after an agent turn settles (inline ghost, below-editor widget, or both).
-
Weather widget for Pi (/weather)
-
Render a document with pandoc and open it in the browser, from inside pi.
-
Token reduction & persistent memory for Pi coding agent — compress tool outputs, save decisions across sessions, survive compaction, backed by Obsidian
-
Register local reference codebases for Pi agent exploration.
-
基于浏览器的代码审查和 Markdown 标注(/pnr /pna /pnl)
-
Hybrid Subagents for Pi: role-based, multi-model specialists powered by pi-subagents