pi-advisor-flow
Advanced Executor/Advisor flow for Pi, fully configurable and extendable.
$ pi install npm:pi-advisor-flowExtensions, skills, prompt templates, and themes published to npm. Install with pi install npm:<package>. See the package docs for details.
Advanced Executor/Advisor flow for Pi, fully configurable and extendable.
$ pi install npm:pi-advisor-flowAgent workflow skills for pi: one entry skill per task type, each with playbooks, over a shared Align → Anchor → Act → Verify loop, plus loop extensions (loop book, anchor gate, decision trail tool, Jev second thought, skill validator, stream rules).
$ pi install npm:pi-anchorloopMakes the Pi agent follow your project's rules. Jev judges every write against your pi-warden.md and quotes the broken rule back to the agent, names slop, breaks stuck loops, calls out unverified done claims, compresses large tool output, and holds the ra
$ pi install npm:pi-wardenPi Coding Agent extension that replaces LLM-summary compaction with TypeSafe Jev keep/drop/truncate decisions.
$ pi install npm:pi-jev-compactionPi extension for building, running and evaluating AgentRun workflows
$ pi install npm:@parcha/agentrun-piTypeSafe AI (Jev) decisions for Pi: batched Choice/Score/Noul evaluation tool, terminal playground, and a typed API other extensions build on.
$ pi install npm:pi-typesafeOne search tool for public web/code, URL extraction, and local evidence continuation. Cited, context-efficient, and keyless by default.
$ pi install npm:@gagansd/pi-web-searchJev, TypeSafe's System One classifier, as a tool inside Claude Code, Codex, Pi, and OpenCode: typed classify, check, score, rank, and ask, plus one-command setup.
$ pi install npm:@french-castle/jev-codePi extension that uses decision/classifier models (Pi, TypeSafe JEV, OpenAI, OpenRouter) to filter which skills are advertised and, opt-in, which external tools and MCP servers are available for each prompt
$ pi install npm:pi-context-decisionsSkill registry for pi — per-project skill on/off, a skill vault, and jev-judged skill exposure
$ pi install npm:@pi-unipi/skill-registryModes and model routing for Pi: every prompt runs as chat, suggest, review, plan or code on the right tier, with plan-then-code built in. Provider-agnostic.
$ pi install npm:@ayhanblt/ay-piJev watchdog for long-running tool calls and background tasks — judges stuck work via TypeSafe System One and kills or warns
$ pi install npm:@pi-unipi/watchdogA minimal permission gate for Pi, inspired by Claude Code's auto mode
$ pi install npm:pi-verdictGreenlight: System One classifiers (TypeSafe Jev and others, via TypeSafe, OpenRouter, Cloudflare and more) as a decision layer for Pi: judges mutating tool calls before they run, reads tool output for secrets and failure class after, escalates flagged ca
$ pi install npm:@khaentertainment/pi-greenlightConfig-driven typed classification for Pi via Microsoft Decision-1, OpenAI Decisions, Perplexity, TypeSafe Jev and Cloudflare Clef, with selective fallback and explicit comparisons.
$ pi install npm:@abhishek944/pi-classifierMy oh-my-pi (omp) harness in one command: a curated collection of public plugins and skills, my Jev router, and switchable Claude / ChatGPT / mixed model setups.
$ pi install npm:omp-maestroAn AI-tone check for pi that runs only while the agent drafts prose: arm a file with prose_watch, and every write or edit to it is judged by Jev, with a likely-AI verdict attached to the tool result
$ pi install npm:pi-prose-watchPi extension that adds MiniMax, StepFun, and Command Code model providers plus a TypeSafe Jev decision tool
$ pi install npm:@pi-vault/pi-providersPer-turn model routing for Pi: a small judge picks the cheap or the strong tier for each message, with multi-model failover, task-level orchestration, and an optional decision-model judge (Jev) that answers with a calibrated probability instead of prose.
$ pi install npm:pi-shift-routerScan Agent Skills for prompt injection, hidden instructions, exfiltration, and remote code before Claude Code, Codex, OpenCode, Pi, or npx skills installs them. Standalone and offline by default; TypeSafe's jev is an optional second opinion.
$ pi install npm:@french-castle/skill-scannerJev virtual model routing for pi with fixed, guarded adaptive and record strategies.
$ pi install npm:@caph42/pi-jev-routerUltra-lean supervisor. Three-layer guardrails. Semantic gate for Pi. Blocks destructive commands offline, checks tests, gates diffs through TypeSafe Jev, OpenRouter, or Cloudflare Workers AI (Clef), and reports Jev/Clef spend in /jev-eye usage.
$ pi install npm:@bismawy/pi-jev-eyeMulti-model dynamic router for Pi Coding Agent powered by native decision classifiers
$ pi install npm:pi-fusion-routerPi extension: let a classifier model pick the thinking level for every prompt, so mechanical prompts stay cheap and hard ones get real reasoning — without switching models.
$ pi install npm:pi-autothinkPi 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 install npm:pi-context-guard-jevTool guardrail for pi with Jev, OpenAI and OpenRouter Decisions providers
$ pi install npm:@techs/pi-decisionSelect text in Pi fullscreen, press Alt+Q, and reply below a neatly formatted quote. No clipboard needed.
$ pi install npm:pi-quote-jevPi extension: filters large tool outputs (text, images, PDFs) down to what is relevant to the agent's task, using TypeSafe's Jev.
$ pi install npm:pi-out-classifyTypeScript-native System 1 decision engine: non-autoregressive choice/score/noul decisions with calibrated probabilities, conformal guarantees, multilingual routing. A Jev-style, laya-compatible open alternative (not affiliated with TypeSafe). Zero Python
$ pi install npm:pi-reflexClassifier-driven model router for pi and Oh My Pi (omp): routes each prompt to a model using Jev (TypeSafe) or Laya typed decisions, with circuit breaking and fallback chains
$ pi install npm:@ejstembler/pi-classifier-routerSingle-surface TypeSafe Jev model router for the Pi coding agent: one virtual model (jev/auto) with judgment routing, mid-turn transient fallback, permanent-failure parking, and first-class routing profiles.
$ pi install npm:pi-jev-router-proSemantic tool routing and typed System One decisions for the Pi coding agent using TypeSafe Jev
$ pi install npm:@leftytennis/pi-jevRoute pi prompts to task-appropriate model tiers with TypeSafe Jev typed judgments. Budget-aware, with automatic fallback.
$ pi install npm:pi-jev-model-routerVerbatim context pruning for Pi using TypeSafe Jev decisions.
$ pi install npm:@dereknex/pi-jev-compactionAuto-names Pi sessions and suggests multi-style titles with /titlewise. Picks with Jev when configured, falls back to the first valid candidate otherwise.
$ pi install npm:@battlelamb/pi-titlewiseGUARDRAILS.md gate for opencode, pi and Claude Code (hooks module): a System One decision model (TypeSafe Jev) judges every bash command before it runs — destructive / credentials / GUARDRAILS.md policy
$ pi install npm:@bergetai/guardrails-mdSemantic desktop observation and goal-driven interaction for Linux applications from the Pi coding agent, over AT-SPI, with in-process TypeSafe Jev System One decisions
$ pi install npm:pi-omarchy-cuRoute each coding-agent turn to the cheapest model that can finish it. Switchyard-style execution signals, TypeSafe's jev as the judge, and honest cost accounting. Works with Claude Code, Codex, OpenCode, and Pi.
$ pi install npm:@french-castle/jev-routerPi extension that uses TypeSafe Jev to route each request into bounded verification, decomposition, or normal response policy.
$ pi install npm:@mohan-cao/pi-jev-response-routerExperimental Codex Sky desktop tools: native MCP for Claude Code, plus pi native/optional Jev modes
$ pi install npm:jev-codex-cuaSystem One decisions for Pi. Plug in TypeSafe Jev, Reflex, or your own provider.
$ pi install npm:pi-system-oneCommunity fork of Luke Parke's pi-subagent with Jev model routing, local tool activation and verified Pi child capabilities
$ pi install npm:@cr1ms0n/pi-subagentPi extension: classify images independently with TypeSafe's Jev (choice, score, yes/no and bounding-box questions).
$ pi install npm:pi-jevisualJev validation guard for pi: on-demand tool + automatic provider gate. Dual backend: TypeSafe direct + OpenRouter.
$ pi install npm:pi-jev-guard