Vision

A persistent, policy-guarded Playwright browser for AI agents with network controls, trusted credential filling, proof screenshots, and CAPTCHA helpers.

MCP server that gives AI agents the web: fetch, search, crawl and screenshot from one local Rust binary. Zero API keys, zero accounts, Chrome-true TLS.

Pi extension for image generation via OpenAI gpt-image, Google Nano Banana (Gemini), Alibaba Qwen-Image, OpenRouter, and custom providers.

Run AI coding agents in pi TUI overlays with interactive, hands-free, and dispatch supervision

Pi package that adds document_parse, document_search, document_screenshot, and a companion skill for local document understanding with LiteParse v2.

Improve OpenAI in pi with fast mode, usage stats, realtime voice, image generation, and footer polish.

Turn an unfamiliar codebase into a validated reimplementation spec, then synthesize confirmed specs and a product vision into a traceable plan.

Pi extension for AI video generation plus local video composition: lossless clip concat and mixed image/video timelines with overlays, TTS, soft or burned subtitles, source audio, BGM, and bundled LGPL/GPL FFmpeg runtimes.

Image generation and image recognition tools for the Pi coding agent

Give text-only pi models vision — describe images with a vision model you pick via an interactive picker, then hand off the text description to non-vision models

Clipboard image paste for pi-archimedes

Linux desktop-control MCP server: AT-SPI accessibility trees, Wayland/X11 input, screenshots, and compositor window targeting.

Automatic image, video and audio description for any model in Pi. Routes media to a multimodal model and injects descriptions into context.

Enhanced read for pi – office docs (docx/pdf/pptx/xlsx etc.) via anydoc, header-only, OCR for scanned pages (Firecrawl or local rapidocr, in the order you configure).

Capability-aware vision + paste extension for the pi coding agent. Delegates image analysis to a vision model only when the active primary model is text-only; passes images through natively for multimodal models (zero delegation).

Paste screenshots into the Pi coding agent without pasting file paths — stable [Image #N] references, clickable history, and compact 480px model thumbnails

Image generation and editing for Pi using your ChatGPT Codex login.

Image and file attachments for Pi — converts pasted/dropped file paths into real image attachments, collapses large text pastes into readable file references, inlines text files, and pastes clipboard file references.

Browser control extension for pi — navigate, click, type, screenshot, and extract data from real Chrome via CDP

Render LaTeX as terminal images in Pi's TUI.

Pi extension for AI multimodal generation and media processing via the multix CLI: images, video, speech, music, 3D, document conversion, and ffmpeg/ImageMagick optimization.

Kepos bridge image generation extension for Pi

Bridge a live pi coding session to Discord: share current work as a digest and keep going from chat. Text and image prompts, one progress card per turn, no bot SDK dependencies.

Preview pasted images in PI Agent before sending them, with inline thumbnails and a compact attachment gallery.

Image attachment and rendering extension for Pi TUI

Register the FreeLLM API gateway (freeapi.n.cofire.cn) as an OpenAI-compatible provider in pi, with automatic model discovery and tools for embeddings, image/video generation, speech, and transcription

Registers a describe_image tool that routes image analysis to a configured vision model through pi's official pipeline.

Pi extension: describes images with a configured vision model when the active model cannot see

Unpublished pi-codex-imagegen workspace for the staged Codex split

Image and optional sprite-sheet generation for Pi via ChatGPT Codex subscriptions, OpenAI, Gemini, Qwen, Seedream, Meta Muse Image, OpenRouter, and custom providers.

Give text-only pi models vision — describe images with a vision model you pick via an interactive picker, then hand off the text description to non-vision models

Ask LLM review, comparison, brainstorming, image, verification, and pairing workflows for Claude Code, Cursor Agent, and Pi

Generate, edit, and analyze images in pi using Google Nano Banana (image gen) and Gemini Vision (analysis). Inline terminal preview, reference-image editing, auto-save.

pi extension: /shake tools|images|thinking — rebuilds session history in place to free context space

Expose Codex subscription text, vision, image, search, Fast, and usage APIs to Pi

Minimal Codex/OpenAI native tools for Pi: Codex image_generation, view_image, apply_patch

Pi extension that turns pasted image paths into first-class image attachments.

Pi quality-of-life extension: compact statusline/π prompt, reliable multiline input, styled pasted-image chips, session naming/search/context import, scheduled prompts, handoff, permission prompts, notifications, custom compaction, and a collapsed-thinkin

Pi Agent extension that adds a describe_image tool, letting non-multimodal models delegate image analysis to a vision-capable model (like Qwen VL)

Codex-compatible native image attachment tool for Pi

One-command installer for the Ai Applied pi stack (pi-vigilant, pi-aia-asf, pi-aia-browser, pi-smart-web-search, pi-smart-fetch, pi-vision-handoff, pi-intercom, pi-safe-compact, compaction-fix).

Pi extension exposing a gpt-5.5+ image generation tool backed by gpt-image-2.5.

Radius Google Workspace and image generation extensions for Pi

TypeScript port of the nano-banana-imagegen pi skill. Generate and edit images with Google Gemini image models via the @the-focus-ai/nano-banana CLI, with GEMINI_API_KEY resolution, output-path handling and batch generation. Exposed as a pi skill and the

Connect pi to 9Router — multi-provider chat models plus image, speech, search, and fetch tools

Give DeepSeek (text-only) models vision in Claude Code, Codex, and Agent Plugins clients: describe images via any OpenAI-compatible vision endpoint.

Paste an image in Pi TUI; relay it with recent chat context and your prompt to a vision model, and inject the analysis into the main model's context

Agnes AI for pi: /model text model catalog (intl + CN) plus image/video generation as callable tools + auto-loaded skill. No model switch needed for media.

A Jupyter-shaped Python notebook for the pi agent: an ordered list of cells over one persistent namespace, with staleness hints, percent-format persistence and plots the model can see.

Give DeepSeek (text-only) models vision in Claude Code, Codex, and Agent Plugins clients: describe images via any OpenAI-compatible vision endpoint.

Pi TUI configuration and an on-demand Skill/CLI for image generation.

Enable Kitty image protocol for localterm in pi's TUI

Vega-Lite chart extension for pi coding agent - render data visualizations as inline images

Readable LaTeX in Pi: terminal-native inline math and MathJax display images.

Generate and edit images using Codex CLI

Codex-compatible native image attachment tool for Pi

Workspace-safe hard-subtitle OCR through the local SubVision macOS service for Pi

Pi Coding Agent extension for OmniRoute — view combos, browse providers, and sync models with enriched metadata (context windows, max tokens, reasoning, and vision) to the Ctrl+P picker

Pi extension package for direct HTTP GET plus web_search, web_fetch, and web_image via Firecrawl, Exa, Tavily, and Brave.

Cynos universal search, vision, and browser tools for the pi coding agent.

Fill any standard Word .docx template — inject AI-authored sections, images, and tables from a template bundle. Generalizes the former manual-creator; controls manuals are one template.

Synthetic (synthetic.new) model provider for pi - Dynamic model fetching with reasoning, vision, and tools support

Local hybrid RAG pipeline for the Pi coding agent. SQLite FTS5 + sqlite-vec, ONNX embeddings via Transformers.js, PDF/DOCX/HTML extraction (with OCR fallback), per-project storage, auto-injection. Zero cloud dependency.

Pi TUI clipboard image attachment bridge

pi coding agent extension for OpenCodeReview (ocr): /ocr-review commands and ocr_review / ocr_delegate tools for AI-powered code review of Git changes.

Socratic planning and shared-understanding sessions for pi.

Image generation and reference-image editing for pi.

Pi extension that resizes oversize images at Read-time so they fit model byte and pixel ceilings.

Windows image picker for pi running in WSL or Windows

Hound web research for Pi agent - keyless search, anti-bot fetch, deep crawl, screenshot, Internet Archive dead-link recovery. Free, MIT, no API key.

Unofficial pi package that exposes Z.ai MCP server tools for web search, URL reading, repository reading, and vision workflows.

Encrypted Bark notifications and image paste placeholders for Pi

Pi extension + skill for a ground→contract→mockup→test→fix→learn frontend design loop. Ships a live mockup server tool, a Playwright breakpoint-screenshot tool, and a design-contract scaffolder. Works in any React/Tailwind/shadcn project.

PDF manipulation, processing, and management toolkit for Pi coding agent — extract text/tables, merge/split, fill forms, create PDFs, OCR, watermark, encrypt/decrypt, and more

Agnes AI provider for pi — registers agnes (token billing) and agnes-plan (subscription) providers with text and image input models

Progressive exact-revision project graph reviews for Pi and Claude Code

MCP vision bridge for text-only coding agents

A Pi extension that delegates current-turn image understanding to a separately configured OpenAI-compatible Vision Provider.

Pi extension that gives non-vision GLM models (z.ai) image understanding via GLM-4.6V

Screenshot picker extension for pi coding agent - quickly select and attach screenshots to your prompts

Thin Pi subagent primitive: one synchronous call for isolated context, live progress, timeout supervision, resumable sessions, and per-task model/tool control.

Automatic jj revision management — guards file edits to keep Jujutsu revisions focused

Pi extension: Zero-setup multi-backend OCR — MinerU (free cloud), Ollama (local GPU, LaTeX formulas), Pix2Text (local Python). Extract text, formulas, and tables from images and PDFs. Default: zero config, works out of the box.

Pi extension: relay tool-result images to Gemini-family user attachments across Responses proxies.

Generate or edit PNG images in Pi using a ChatGPT Plus/Pro Codex subscription.

DeepSeek vision image offload for pi: Files API file_id references on the official gateway (deepseek-v4-flash-vision-exp), offload trimming on third-party gateways — kills the 50 MB 413 ceiling.

Build knowledge artifacts that survive revision and session boundaries — plans, specs, repo or data analyses, reports — with declarative seams, a checkpoint, provenance, and a domain-agnostic patch tool.

pi-web plugin that browses and previews screenshots saved by pi-web into .pi-web/paste/ — gallery panel, lightbox, and chat-inline previews for folder-mode attachments.

ComfyUI image/video generation extension for pi coding agent

Pi extension: let a text-only model read images by asking a vision-capable model from your own models.json.

Log in to the airpx LLM proxy (https://airpx.cc) with an sk-proxy API key and auto-import its model catalog (prices, context windows, reasoning, vision) into the pi coding agent.

pi extension + skill: visual understanding via a configurable vision model (pi-registry reuse or custom responses API; default gpt-5.6-luna) when the main model cannot see images — pure TypeScript, single runtime

Transparent vision fallback for text-only pi models — overrides read so image files are described by a vision model when the active model cannot see images. Supports any OpenAI-compatible endpoint.

GPU infrastructure control for Pi: provision, deploy, run, monitor, pause, and destroy GPU video workloads on RunPod (the one first-class adapter; more plug in via the adapter pattern) — MiniMax H3 on RTX PRO 6000 with the Turbo 8-step / SageAttention2 /

Archived and no longer maintained: Code Reasoning tools for pi and MCP — reflective sequential thinking with branching and revision support

Give text-only pi models vision — a native vision tool with a bundled vision script and a multi-backend model chain (auto-fallback, timeout, retry). Configure models via .env, JSON, or a custom script.

Pi agent extension that parses image/audio/video inputs and forwards them to an OpenAI-compatible multimodal proxy (e.g. dots.ai).

Local MLX vision for text-only models on Apple Silicon Macs: a describe_image tool backed by LiquidAI LFM2.5-VL-3B-MLX-8bit, so non-vision models (DeepSeek, etc.) can see images.

GPU infrastructure control for Pi: provision, deploy, run, monitor, and destroy GPU workloads on RunPod, Vast.ai, SaladCloud, Modal, Lambda, and Prime — with H3/Wan/Hunyuan video generation as the first-class workload. First-run onboarding, spend policy,

Browser automation and web research tools for pi — search, visit, screenshot, interact, and inspect console output

Muted thinking blocks, framed editor, animated header, response time, rich footer, and clipboard image paste for pi

Pi extension for image generation: Codex (ChatGPT OAuth via pi, no Codex CLI) plus any OpenAI-compatible provider (llm-center, OpenAI, proxies).

An issue-tracker extension for pi that turns goals into linked, tracked user stories stored in a project-local SQLite database.

Convert any document to Markdown with Microsoft's markitdown CLI — PDF, DOCX, PPTX, XLSX, images (OCR), audio (transcription), YouTube, HTML, CSV, JSON, XML, EPUB, ZIP

Smart clipboard paste for pi on Windows — copy files and paste their paths; falls back to images.

pi 视觉理解桥接扩展：当前模型不支持图片时，自动调用外部视觉模型分析图片并注入结果

Paste images from the Mac clipboard into pi running on a remote machine over ssh. A launchd socket service on the Mac serves the clipboard image; an ssh RemoteForward carries it; the extension fetches it on ctrl+v and attaches it like a native paste.

Paste Windows clipboard text and images into Pi from WSL with Ctrl+V.

Find skipped Mermaid diagrams and show them in a terminal image viewer.

Codex image generation and editing for Pi, Code Mode and Notebook Mode.

Pi skill — a mechanical, countable anti-slop checklist for AI-generated frontend. Catches the specific tells an undirected model defaults to (AI-purple, Inter-everywhere, em-dashes, div-based fake screenshots, eyebrow-per-section, Jane Doe / Acme data). A

Pi extension for web search, URL content extraction, and image search (Tavily, Kimi, DeepSeek, Mimo, Z.AI, DashScope, Unsplash, and more)

Asynchronous Pi subagents and agent fleet orchestration for parallel coding agents with nested delegation, background work, and supervision in herdr.

Transparent image-to-text bridge for text-only Pi models using a configurable vision model

OpenAI toolkit for Pi: Codex Remote Context windows, remote compaction v2, routed Web Search, image gen, auto mode.

Graphviz chart extension for pi coding agent - render DOT diagrams as inline images

Browser automation tool for pi — interactive browsing, screenshots with inline vision, and session cleanup via agent-browser CLI

Pi extension for web search, page extraction, Firecrawl scraping/crawling, Crawl4AI headless browser crawling, real-browser interaction (trusted click/type/evaluate via CDP), Gemini web-tier research, free upstream image generation (Gemini/ChatGPT web/Z.a

Find skipped Mermaid diagrams and show them in a terminal image viewer.

Essential extensions for pi — screenshots, image context pruning, a daily log tool, and a markdown viewer.

Temporary pasted-image cache with compact placeholders for the pi coding agent.

QwenCloud provider for pi — access Qwen3.8, DeepSeek V4, GLM-5.2, Wan image generation, and HappyHorse video generation through QwenCloud's OpenAI-compatible API

Pi-native ralph loop — autonomous coding iterations with mid-turn supervision

Pi package for OpenAI/Codex image generation with a local browser studio.

A cute, colorful pet companion for the pi coding agent. Use classic ASCII pets or Petdex image pets that live below your editor and react to tools, tests, commits, PRs, reviews, subagents, model swaps and more.

Image preview extension for pi coding agent — renders inline image thumbnails above the editor using kitty graphics protocol with tmux support

Pi extension for literature search, Zotero, PDF parsing, citations, materials data, and scientific image generation

Enhanced Qwen Token Plan CN provider for pi. Speaks the OpenAI Responses API (not chat-completions) to activate the platform's server-side Harness tools — web search, code interpreter, web extraction, and image search — that the built-in qwen-token-plan-c

Translate EPUB books in Pi while preserving structure, styling, images, footnotes, and packaging.

OCR image-to-text tool for Pi — extracts text from screenshots, terminal output, and code images using Tesseract + ImageMagick

OpenRouter multimodal tools for Pi — search, fetch, image gen, vision, video, PDF, TTS, STT

Typed Pi tools for browser automation, rendered-page context extraction, screenshots, and saved session state via agent-browser.

OpenCode-grade clipboard paste for pi on Windows: instant Ctrl+V image attachments, Explorer file drops as collapsible blocks, large-text collapsing, persistent clipboard helper

Standalone Pi extension for OpenAI fast mode, GPT-5.6 Pro request injection, and image generation/editing.

Pi extension for NVIDIA NIM — chat, vision (VLM), embeddings, and image generation

Interactive Playwright web browsing for Pi. Chromium and Firefox built in; author stealth backends on either engine. Snapshots and screenshots cache to disk, saving context. Configurable profiles/cookies. Agents write site guides, auto-matched by domain.

DeepInfra provider for pi: dynamic model catalog, reasoning-effort thinking levels, vision, session usage + monthly billing footer

The terminal previewer for pi — tabbed side drawer for markdown, syntax-highlighted code, images, and PDFs (text + visual pages).

Atomic image placeholders for pi 0.85.0 on macOS

Multimodal perception for Pi — image/audio/video/document understanding, transcription & image generation via Gemini (gemini_api or local antigravity CLI).

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.

Pi extension for Cloudflare Workers AI — chat, image generation, TTS, and speech-to-text over the free tier

Pi extension for D.AI image, video, and music generation

Generate and edit images in Pi using existing Codex and Grok subscription account quotas.

Model-profiled Codex Responses tools for Pi: WebSocket, Lite, web search, images, compaction, and apply_patch

Display images inline in the pi coding agent: native protocol (iTerm2/Kitty/WezTerm/Ghostty), Sixel in Windows Terminal, ANSI half-blocks everywhere else (works under Herdr/tmux).

Pi extension that turns pasted image paths into first-class image attachments with automatic optimization.

Lean Z.ai tools for pi: URL reader, GitHub zread lookup, and image/video vision. No search tool (use pi-web-access). Minimal token footprint.

Enable Kitty image protocol for Warp terminal in pi's TUI

Pi extension for GMI Cloud Inference Engine — OpenAI-compatible chat, vision (VLM), and embeddings

Ignition 8.3 gateway + project assistant — file-based resource authoring, REST scan/hot-reload, tag scaffolding, screenshot-verified Perspective UI

Unified pi extension for a self-hosted Lemonade server: dynamic chat-model discovery, multimodal agent tools (transcription, image gen/edit/upscale, TTS, audio, 3D mesh), a live /lemonade-setup management TUI, and a lemonade-only below-editor status bar (

Native pi agent image generation and editing through OpenAI-compatible Images APIs.

Pi skill + subagent: three parallel code-review channels (OCR / Standards / Spec) reviewing the same diff bound under one scope, then an oracle deduplicates, resolves conflicts and re-grades them into a single adjudicated report (00-adjudication.md). Pure

Browser automation skill — drive your real Chrome via the browser-use CDP CLI (navigate, screenshot, coordinate-click, extract, tabs). Penetrates closed shadow-DOM & cross-origin iframes via compositor-level coordinate clicks where js selectors fail.

Pi extension: auto-attaches pasted/dragged image file paths as ImageContent blocks so the model actually sees the image.

Image gallery, remote URL caching, multimodal delivery, and macOS Quick Look for Pi

Image generation and editing for Pi Agent through the Zhizhengroup LLM Gateway (generate_image tool)

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: Z.AI GLM-4.6V vision tools — image analysis, OCR, error diagnosis, diagram reading, UI diff, UI-to-code, video analysis

Tmux-safe, TUI-only local image previews for Pi

Pi extension: review and test JS frontends headlessly — open dev-server pages in Playwright Chromium, interact with them, capture screenshots the model can see, and collect console/network errors.