Package Catalog
Extensions, skills, prompt templates, and themes published to npm. Install with pi install npm:<package>. See the package docs for details.
Recently published
All packages
pi-deepseek-cache
DeepSeek prefix-cache extension for pi — hit-rate telemetry, prefix guard, and cache-friendly compaction.
$ pi install npm:pi-deepseek-cache@kennyfrc/pi-deepseek-minimal-mode
Strict-by-default DeepSeek Harness composition for Pi, with an opt-in augmented profile that inlines configured extra tool definitions and durable AGENTS context.
$ pi install npm:@kennyfrc/pi-deepseek-minimal-modepi-relace-provider
Register Relace (models.relace.ai) as an OpenAI-compatible model provider for pi
$ pi install npm:pi-relace-providerpi-deepseek-cost
DeepSeek usage & cost tracker for Pi — live session cost in the status bar, /ds-cost overlay panel, bilingual UI with CNY/USD/EUR display, official peak/off-peak billing
$ pi install npm:pi-deepseek-cost@juanmackie/pi-deepseek-peak
Show DeepSeek scheduled peak/off-peak pricing and API account health in the pi status bar.
$ pi install npm:@juanmackie/pi-deepseek-peakpi-peak-hours
Peak/off-peak billing-hours indicator for LLM APIs (DeepSeek, GLM Coding Plan) in the pi coding agent.
$ pi install npm:pi-peak-hourspi-dsh-mimic
Unlock DeepSeek V4 Pro's DSH Minimal capability in Pi while keeping Pi's full plugin ecosystem
$ pi install npm:pi-dsh-mimicpi-check-agent-quota
Show AI provider quota and balance in the pi TUI status widget
$ pi install npm:pi-check-agent-quotapi-inferx-provider
Pi coding-agent extension that registers InferX (https://model.inferx.net) as a model provider — live /v1/models discovery with a disk cache, a built-in public catalog fallback, and vLLM compatibility shims (system role, max_tokens, Qwen/DeepSeek/GLM thin
$ pi install npm:pi-inferx-providerpi-provider-status
Show remaining quota and balance for LLM providers in pi — /status panel for Kimi Coding, Zhipu ZAI, MiniMax, DeepSeek, OpenRouter, Claude and Codex subscription windows
$ pi install npm:pi-provider-statuspi-no-spin
pi extension that detects when the LLM spins (repeats the same string over and over), interrupts generation, and reminds it to stop
$ pi install npm:pi-no-spin@amartinr/pi-bifrost-reasoning-fix
Pi extension that fixes the DeepSeek-behind-Bifrost reasoning drop on tool-calling turns. Bifrost relays DeepSeek reasoning in non-standard fields (reasoning / reasoning_details instead of reasoning_content); in tool-calling contexts DeepSeek requires rea
$ pi install npm:@amartinr/pi-bifrost-reasoning-fix@wuxiangru/pi-vision-bridge
Give text-only models (DeepSeek, Llama, etc.) vision — describe_image tool + automatic image-to-text fallback, using any vision model configured in pi's model registry.
$ pi install npm:@wuxiangru/pi-vision-bridgepi-dsh-pet
Floating desktop deepseek chan for the pi coding agent: idle breathing, random actions, screen wandering, and pi event reactivity (thinking/coding/idle → matching animations).
$ pi install npm:pi-dsh-petpi-deepseek-harness-anchor
DeepSeek V4 Pro/Flash harness trajectory anchoring for Pi: Minimal-persona first turn, narrow tool pair, durable promotion, DSH-compatible str_replace_editor
$ pi install npm:pi-deepseek-harness-anchor@limccn/deepseek-vl-support
Give DeepSeek (text-only) models vision in Claude Code, Codex, and Agent Plugins clients: describe images via any OpenAI-compatible vision endpoint.
$ pi install npm:@limccn/deepseek-vl-support@kennyfrc/pi-str-replace-editor
Model-facing str_replace_editor (view/create/str_replace/insert) for DeepSeek-family models on Pi; replaces read/edit/write for those models
$ pi install npm:@kennyfrc/pi-str-replace-editor@99percentpeople/pi-deepseek-anchor
Simulate DSH minimal mode's first request to induce a stronger DeepSeek V4 Pro agent trajectory
$ pi install npm:@99percentpeople/pi-deepseek-anchorpi-anchored-tool-for-dspro
Pi extension that anchors a target model's first request to a minimal tool catalog (shell + read), then restores the full catalog after the first tool call. For DeepSeek V4 Pro and other models that condition on the API-visible tool list (Standard/PTC 91/
$ pi install npm:pi-anchored-tool-for-dspro@aiwayds/pi-bailian-token-plan
阿里云百炼 Token Plan 扩展 - 支持千问、DeepSeek、GLM 等模型
$ pi install npm:@aiwayds/pi-bailian-token-planpi-opencode-go-cache
Prompt caching for Pi's OpenCode Go provider (kimi, deepseek, mimo, qwen, minimax). Stamps prompt_cache_key, 24h retention, and cache_control breakpoints on every request — beyond what pi-ai or opencode CLI do by default.
$ pi install npm:pi-opencode-go-cachedeepseek-peak-hours-warning
Pi extension that warns during DeepSeek peak billing hours and locally applies the active peak/off-peak rates to DeepSeek model cost estimates.
$ pi install npm:deepseek-peak-hours-warningpi-deepseek-router
Task-aware DeepSeek routing extension for Pi with strict no-op for other models.
$ pi install npm:pi-deepseek-routeraegis-method-pack
Aegis Method Pack: make your AI coding agent trustworthy — fewer reworks, safer changes, proof before done. Multi-harness method pack (Claude Code, Codex, OpenCode, Cursor, Windsurf, CodeBuddy, DeepSeek Harness, Kimi, and more).
$ pi install npm:aegis-method-packpi-deepseek-peak-alert
Pi extension that warns you with desktop notifications when using DeepSeek during peak pricing hours (2× rates)
$ pi install npm:pi-deepseek-peak-alert@mcowger/pi-llmgateway
A Pi extension that adds [LLM Gateway](https://llmgateway.io) as a model provider, giving you access to 160+ chat models from OpenAI, Anthropic, Google, xAI, Mistral, DeepSeek, and many others through a single OpenAI-compatible API.
$ pi install npm:@mcowger/pi-llmgatewaypi-deepseek-pet-extension
Non-blocking Pi lifecycle status bridge for Pi DeepSeek Pet
$ pi install npm:pi-deepseek-pet-extensionpi-turn-metrics
Pi Coding Agent extension for real-time execution turns, steps, LLM/tool durations, TTFT, and TPS in the footer status bar
$ pi install npm:pi-turn-metricspi-deepseek-anchor
pi extension: DeepSeek V4 Pro Minimal-anchored bootstrap, then full Standard tools (port of dsh-anchored-standard)
$ pi install npm:pi-deepseek-anchor@yankewei/pi-web-search
pi extension that exposes provider-native web search for DeepSeek Responses and OpenAI Codex GPT-5.6 models
$ pi install npm:@yankewei/pi-web-search@jours_757/pi-extensions
Pi 扩展合集：当前包含 vision-bridge（识图桥接——自动把图片识别交给视觉模型，文字描述交回文本主模型）。
$ pi install npm:@jours_757/pi-extensions@jetserge/pi-usage-bars
Quota, balance, and spend indicators for current Pi providers
$ pi install npm:@jetserge/pi-usage-barspi-byteplus-modelark
Pi extension for BytePlus ModelArk Coding Plan — registers ModelArk as a custom OpenAI-compatible provider with all supported coding models.
$ pi install npm:pi-byteplus-modelark@oversk7/pi-web-dsh-style
An unofficial DSH-styled Windows browser UI for pi coding agent with configurable LAN and native frp relay access
$ pi install npm:@oversk7/pi-web-dsh-style@demo-0416/pi-trace
pi extension: deepseek-harness-style trajectory view — timeline gantt, turn-grouped ledger, rich detail panel, live SSE streaming
$ pi install npm:@demo-0416/pi-tracepi-qwencloud-provider
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 install npm:pi-qwencloud-providerpi-provider-alibaba-complete
Alibaba Cloud provider for Pi with Coding Plan, Token Plan, pay-as-you-go, dynamic catalogs, and verified reasoning controls
$ pi install npm:pi-provider-alibaba-completeyimo-pi-kit
A portable Pi coding-agent profile with extensions, subagents, skills, prompts, themes, DeepSeek Responses Web Search, and optional local code-review graph/MCP integration.
$ pi install npm:yimo-pi-kitpi-vision-native
Native vision for text-only pi models (DeepSeek, Llama, ...). Capability-aware chat attachment interception + vision-describe/vision-ask tools via any OpenAI-compatible vision endpoint (ModelScope, SiliconFlow, OpenRouter, GLM, Ollama), with intent detect
$ pi install npm:pi-vision-nativepi-deepseek-v4-anchor
Two-phase DeepSeek V4 Pro tool bootstrap for pi: Minimal-anchored first request (bash + str_replace_editor + DSH persona), then full catalog. Port of xiaobright/dsh-anchored-standard.
$ pi install npm:pi-deepseek-v4-anchorpi-aws-bedrock-mantle
Pi extension: all Bedrock Mantle models (GPT-5.5, DeepSeek, Qwen3, Mistral, Kimi, ...) via SigV4 — no long-term API key needed
$ pi install npm:pi-aws-bedrock-mantlepi-vision-bridge-tomoyo
Vision bridge for text-only models (e.g. DeepSeek V4 Flash): images pasted into the chat and screenshots are described by a cheap vision model (MiniMax M3 via opencode.go) and fed to the main model as text. Registers a describe_image tool for on-demand sc
$ pi install npm:pi-vision-bridge-tomoyopi-deepseekx
A Pi extension for configurable minimal prompting and neutral bash safeguards.
$ pi install npm:pi-deepseekxpi-deepseek-startup
DeepSeek branded startup UI for Pi: whale splash screen, animated whale working indicator, and a matching DeepSeek theme. Bundles the startup-ui adapter, the DeepSeek page, and the DeepSeek theme in one installable package.
$ pi install npm:pi-deepseek-startuppi-v4-jspace
DeepSeek V4 Minimal Anchor + J-Space runtime integration for Pi
$ pi install npm:pi-v4-jspacepi-deepseek-responses-search
DeepSeek V4 on the Responses API with native server-side web search (OpenCode Zen/Go + DeepSeek first-party) — search markers shown in the thinking chain
$ pi install npm:pi-deepseek-responses-searchpi-usage-plus
Super usage/quota extension for Pi & Pi Web – 10+ providers, Kimi/Moonshot, OpenRouter, OpenCode Go, DeepSeek, GLM, dual Pi & Pi Web compatible
$ pi install npm:pi-usage-plus