Scout — Sourcing

Scout is the governed sourcing & vendor-selection app: compare vendors, promote evidence through Converge gates, and produce decisions that can be audited.

This repository is being migrated from a hackathon starter kit into a product baseline:

• Web app: Svelte/SvelteKit front end, deployed through Firebase Hosting and supported in Firefox.
• Backend: Rust services, with the current HTTP harness evolving toward a gRPC API boundary.
• Database: monitored Google Cloud database for durable decisions, evidence, audit trails, and operational telemetry.
• Desktop app: Tauri 2 + Svelte + Rust, packaged for macOS Apple silicon, macOS Intel, and Windows.

The governance model remains the same: agents propose facts, Converge promotes valid proposals through policy gates, and every decision either converges with evidence or stops honestly.

Product Rules

• Scout (this repo) is the one product truth.
• Supporting runtimes are examples or migration fixtures, not product workflows.
• The repo uses one train: main plus release/<version> branches only.
• Do not use git worktrees.
• Do not create feature branches for normal work.

Stack

Quick Start

just setup
just check
just test
just dev

Start only the backend harness:

just server

Run a governed vendor evaluation:

curl -X POST http://localhost:8080/v1/truths/vendor-selection/execute \
  -H 'Content-Type: application/json' \
  -d '{"inputs":{"vendors":"Acme AI, Beta ML, Gamma LLM"},"persist_projection":true}'

Launch the desktop app:

just desktop

Package the desktop app:

just package-desktop

Project Layout

crates/
  governance-kernel/       Domain model and in-memory store
  governance-truths/       Vendor-selection truth and Converge bindings
  governance-server/       Local backend harness and truth executors
  governance-app/          Shared Rust app layer
apps/
  desktop/                 Tauri 2 + Svelte desktop shell
examples/
  vendor-selection/        Seed data, policies, and example requests
kb/                        Obsidian knowledge base

Key Docs

• AGENTS.md - canonical instructions for all agents
• MILESTONES.md - current product roadmap
• CAPABILITIES.md - what the starter supports
• DEPLOYMENT.md - Google Cloud and Firebase deployment plan
• RELEASE.md - desktop release packaging plan
• CONTRIBUTING.md - contributor workflow
• SECURITY.md - vulnerability reporting
• SUPPORT.md - support channels

Rules

• No unsafe code.
• Use typed enums for semantic state.
• Every mutation needs an Actor.
• Suggestors emit proposals, not direct facts.
• Providers stay behind trait boundaries.
• Run just lint before considering code complete.

License

MIT. See LICENSE.