greedysearch

GreedySearch for Pi

Multi-engine AI web search for Pi via browser automation.

• No API keys
• Real browser results from configurable engines: Perplexity, Google AI, ChatGPT, Gemini, plus opt-in Semantic Scholar and Logically research engines
• Research mode as the centerpiece: iterative planning, source fetching, citation audit, and structured bundles
• Optional configurable synthesis with source grounding (Gemini by default)
• Chrome runs headless by default — no window, purely background

Install

pi install npm:@apmantza/greedysearch-pi

Or from git:

pi install git:github.com/apmantza/GreedySearch-pi

Tool

• greedy_search — multi-engine AI web search, source-grounded synthesis, and deep research

Quick usage

greedy_search({ query: "React 19 changes" }); // all engines + fetched sources
greedy_search({ query: "React 19 changes", synthesize: true }); // add configured synthesis
greedy_search({ query: "Prisma vs Drizzle", engine: "perplexity" }); // individual engine
greedy_search({
  query: "Evaluate browser automation options for AI agents",
  depth: "research",
  breadth: 3,
  iterations: 2,
  maxSources: 8,
});
// Research mode writes a dataroom-style bundle under .pi/greedysearch-research/ by default.
// Headless is the default — no window. To force visible Chrome:
greedy_search({ query: "Visible browser setup", engine: "perplexity", visible: true });

Parameters (greedy_search)

Common

• query (required)
• fullAnswer: return full single-engine output instead of preview
• headless: set to false to show Chrome window (default: true)
• visible / alwaysVisible: set to true to always use visible Chrome for this search

Normal search

• engine: all (default web/search fan-out), perplexity, google, chatgpt, gemini; opt-in research engines: semantic-scholar, logically; bing still works for signed-in users
• synthesize: for engine: "all", synthesize fetched sources with the configured synthesizer (default false)
• synthesizer: override the configured synthesis engine for this call (gemini default, chatgpt supported)
• depth: legacy fast/standard/deep aliases are still accepted for compatibility; prefer synthesize for normal search

Deep research

• depth: set to "research" to run the iterative research workflow
• breadth: number of research actions per round, 1-5 (default 3)
• iterations: research rounds, 1-3 (default 2)
• maxSources: fetched source cap for the final report, 3-12
• researchOutDir: optional directory for the research bundle
• writeResearchBundle: write the research bundle to disk (default true for research mode)

Deep research uses the configured ~/.pi/greedyconfig.engines list for child searches and Gemini for planning/final synthesis.

Environment variables

Variable	Default	Description
GREEDY_SEARCH_VISIBLE	(unset)	Set to 1 to show Chrome window instead of headless
GREEDY_SEARCH_ALWAYS_VISIBLE	(unset)	Set to 1 to force visible Chrome for all GreedySearch runs
GREEDY_SEARCH_IDLE_TIMEOUT_MINUTES	5	Minutes of inactivity before auto-killing GreedySearch Chrome
GREEDY_SEARCH_LOCALE	en	Default result language (en, de, fr, es, ja, etc.)
CHROME_PATH	auto-detected	Path to Chrome/Chromium executable

Search modes

• Individual engine search/research — engine: "perplexity" | "google" | "chatgpt" | "gemini" | "semantic-scholar" | "logically" | "bing"; returns that engine's answer and sources.
• Grounded multi-engine search — default engine: "all"; fans out to configured engines, ranks sources, fetches top source content, and reports confidence metadata.
• All + synthesis — add synthesize: true (or CLI --synthesize) to ask the configured synthesizer to combine engine answers and fetched source evidence.
• Deep research — depth: "research"; iterative action planning, direct URL fetches, fast multi-engine searches, source fetching, learning extraction, deterministic floor checks, citation audit, a final cited report, and a structured on-disk bundle.

Legacy depth: "fast" | "standard" | "deep" values remain accepted for compatibility: fast skips source fetching; standard/deep request synthesis.

Configure all-engine fan-out and synthesis in ~/.pi/greedyconfig:

{
  "engines": ["perplexity", "google", "chatgpt", "gemini"],
  "synthesizer": "gemini"
}