Scout (Second-Brain)
wiki/decisions/0002-universal-acquisition-engine.md
ADR-0002: Universal Acquisition Engine
Date: 2026-06-14 Status: LOCKED Decision: Scout is a standalone universal acquisition engine with dual interface (HTTP API + Python library).
Context
Scout was tightly coupled to PRISM. Other tools (Claude skills, ad-hoc scripts) also needed web acquisition. Should Scout be a PRISM module or a standalone engine?
Decision
Scout is standalone. Two access modes:
1. HTTP API — FastAPI service at localhost:8421. Any language, any consumer.
2. Python library — from scout.core import ScoutCrawler. Direct import for Python consumers.
Consumers import Scout; scout.core never imports a consumer.
Consequences
- Scout has its own repo, its own test suite, its own deployment.
- PRISM calls Scout over HTTP (or imports the library). Scout doesn't know PRISM exists.
- Same engine powers CLI (
scout scrape), API (POST /scrape), and library (ScoutCrawler.scrape()).