Untested on omp
-
Markdown-first collaborative worksheets for Pi
-
Project and session-aware NTFY notifications for the pi coding agent.
-
Persist the last selected model and each model's thinking level to Pi's settings.json.
-
Claude Code-style working spinner for Pi — shimmer, thinking glow, token count, stall detection
-
Set and apply per-model thinking levels from Pi's native settings.
-
Show per-model prices in CNY (in/out/cacheRead/cacheWrite + tiers) inside pi's built-in /model selector
-
Pi extension: manage shared backend processes via the sharedserver CLI.
-
OpenTelemetry traces and metrics for the pi coding harness, exported via OTLP/HTTP to Grafana otel-lgtm
-
Lazy tool loading for pi-web-access.
-
/provider command for pi: add, edit and test custom providers (relays, proxies, local servers) in models.json
-
HiViz — audit and refactor standing AI-agent instructions (AGENTS.md/CLAUDE.md/skills/MCP) for prompt debt. Trace every rule before cutting it: probe-verified, safety-preserving; drift checks with MCP usage telemetry, constitution tests, blame provenance,
-
A minimal Pi loadout menu for project-local packages.
-
Pi skills and prompts for deep multi-lens code review, advisory AI-slop and driver-curation assessment, and PR/branch size analysis.
-
Dedicated-model, safety-first compaction extension for Pi Coding Agent.
-
Mnemos memory & knowledge server — Pi extension. Spawns `mnemos mcp-server` over stdio and exposes every mnemos_* tool as a native Pi tool, plus the mnemos skill pack.
-
Gives Pi a sparring partner to discuss, debate, push back, and fact-check.
-
Opinionated sub-agent router for Pi: tightly couples SKILL.md execution with the Augment Code (auggie) Context Engine.
-
Local inspector for continuation gate diagnostics.
-
Ollama extension for pi-coding-agent. Unified local + cloud Ollama support with model management
-
Pi coding-agent extension that rewrites the editor prompt for GPT or Claude, with shortcuts and settings.
-
Connect pi to Anthropic's Claude Design MCP server: /design-login command plus auth CLI for pi-mcp-adapter
-
Compact long agent sessions earlier, at a finished piece of work, to cut cost without losing results (Claude Code, pi, Codex)
-
Non-blocking, privacy-conscious automatic session names for Pi
-
Pi skill: automatically download Sentinel-1 SLC IW data from ASF for SBAS-InSAR experiments. Same-track, same-direction, per-date full-coverage guarantees.
-
Minimal pi footer: cwd, git branch, model with thinking level, tps/ttft/cache stats, quota, goal, and a context usage bar.
-
MCP client extension for pi
-
InferHub provider with a live model catalog for Pi
-
Jump between user prompts in Pi's fullscreen conversation without changing session history.
-
Prevents concurrent pi sessions from clobbering each other's llama.cpp KV cache — each session is pinned to the best available slot (empty > cold > low tokens > low id) via the server's /slots endpoint, coordinated across windows through a state file
-
Fix malformed single-line Markdown code fences in PI Agent responses so code blocks display correctly.
-
Pi extension package for the [Poolside](https://poolside.ai) inference API.
-
Source-integrated Pi context management, remote compaction v2, image generation, Astra compatibility and apply_patch tools
-
Undo/redo for pi, but this time it works. Per-message snapshots, ported from OpenCode's exact undo/redo philosophy.
-
Pi extension that shows Codex ChatGPT subscription usage without requiring Codex CLI.
-
Web search and fetch extension for pi coding agent
-
Azure DevOps for pi: native read-only tools, a skill, and a zero-dependency `ado` CLI for work items, bugs, WIQL and pull requests.
-
Search local Open Knowledge Format content from Pi with a native Rust/Tantivy backend.
-
Declarative lifecycle rules over an append-only slot ledger, as a pi extension
-
Pi extension: all Bedrock Mantle models (GPT-5.x, Claude, DeepSeek, Qwen3, Mistral, Kimi, ...) via SigV4 — no long-term API key needed
-
Quiet Display for Pi: verb-first tool rows (built-in + Foreign), Verb Groups, sticky on/off.
-
Open lazygit without leaving your current Pi session.
-
Pi extension that registers Melious (melious.ai) as an OpenAI-compatible LLM provider.
-
Pi extension for built-in tools management and prefix-caching-friendly tool gating
-
Mnemos memory & knowledge server — Pi extension. Spawns `mnemos mcp-server` over stdio and exposes every mnemos_* tool as a native Pi tool, plus the mnemos skill pack.
-
Pi provider extension for [Edgee](https://www.edgee.ai), the Agent Gateway that compresses, routes, and observes LLM requests.
-
Explicit, resumable issue-to-MR workflow for the Pi coding agent.
-
Permission enforcement extension for the Pi coding agent.
-
Configurable tool-call display density for pi: mini / low / medium / default. Batch summary lines, compact rows, hover highlight and a settings command.
-
Configurable web search and web fetching for Pi coding agent.
-
Pins your last user message to the top of pi's fullscreen TUI; clicking the pinned bar jumps to that message in the transcript