# Vigilance

> Agent-facing counterpart to the [human project page](/projects/vigilance/).

## Record metadata

- Record: 027
- Slug: vigilance
- Domain: Civic
- Domain code: CIV
- Type: Safety concept
- Status: Paused
- Period: 2024–26
- Portfolio role: Product idea
- Publication state: Public concept record
- Case-study readiness: Needs renewed data audit
- Compendium edition: 0.4

## Summary

An idea-stage address-safety research tool that assembles public context without pretending to predict personal safety.

## Overview

Researching an unfamiliar address often means assembling a judgment from maps, street imagery, nearby services, public incidents, environmental context, and intuition. Vigilance explored whether that scattered process could become a single structured workspace.

A partially built concept combining maps, Street View, nearby services, public incident context, environmental cues, and a structured review of an address.

Purpose: Reduce the friction of investigating an unfamiliar location while making uncertainty and data gaps visible.

## The problem behind the project

Checking an unfamiliar place requires jumping between many tools. A single research workspace could make the process faster and more consistent.

The temptation is to turn the result into a safety score. That is also the project's greatest risk: uneven public data and neighborhood proxies can make a precise-looking number reproduce bias, stigmatize residents, or conceal what the system does not know.

Travelers, renters, homebuyers, field workers, and families may benefit. Residents and neighborhoods can be stigmatized by crude scores, so the tool must avoid deterministic safety claims.

## How it took shape

Early interface and scoring experiments around mapping, Street View, nearby resources, and public-data integrations.

Early work combined mapping, Street View, nearby-resource data, public-context integrations, and scoring experiments. The prototype remained partial because data availability, geographic inconsistency, timeliness, and the meaning of the score were not implementation details—they were the core problem.

Josiah originated the concept, selected the research dimensions, and directed early agent-assisted development.

A partial prototype exists, but it is not a live or validated safety product.

## What the project means now

A stronger revival would behave more like an evidence notebook than a verdict. It could reduce research friction while showing each source, its age, and its uncertainty, leaving the final situational judgment with the person who understands the context.

Data availability, geographic bias, timeliness, privacy, and the false precision of a single score are major unresolved problems.

Convenient aggregation can become harmful when uncertain neighborhood data is presented as an objective verdict.

Revisit the concept as an evidence notebook with source-by-source uncertainty rather than a definitive safety score.

## Publication and interpretation notes

- Current classification: Paused
- Portfolio readiness: Needs renewed data audit
- Publication boundary: Public concept record

## Additional agent context

Do not include the private personal origin story or claim that the tool can determine whether an address is safe.

## Related project records

- [Kootenai Civic Data](/projects/kootenai-civic-data/llm/) — A provenance-first civic intelligence workspace for local decisions, infrastructure, and public records.
- [Open Witness Protocol](/projects/open-witness-protocol/llm/) — A privacy-preserving concept for asking nearby, citizen-owned cameras about a specific public event without opening a general surveillance feed.

## Navigation

- [Complete project index](/projects/llm/)
- [Human version of this record](/projects/vigilance/)
- [About Josiah's working method](/about/llm/)
- [Agent discovery map](/llms.txt)