Menu
Run the full TIGER defect audit pipeline for a MetroNow service zone — fetch from Overpass, classify defects, analyze history, generate reports.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
| name | zone-audit |
| description | Run the full TIGER defect audit pipeline for a MetroNow service zone — fetch from Overpass, classify defects, analyze history, generate reports. |
| when_to_use | User says scan, audit, check a zone, run an audit, or mentions a specific zone name like blue-ash-montgomery |
| allowed-tools | Read Grep Glob Bash(node *) Bash(python *) |
| argument-hint | [zone-name] |
| arguments | ["zone"] |
Run the full audit pipeline for MetroNow zone: $zone
!python -c "import sys; sys.path.insert(0, 'src'); from osm.zones import ZONES; [print(f' {k}: {v[\"name\"]} — {v[\"description\"]}') for k,v in ZONES.items()]"
!found=0; for d in osm-audit-*/; do if [ -f "$d/scan-results.json" ]; then echo " ${d%/} — last scan: $(date -r "$d/scan-results.json")"; found=1; fi; done; [ $found -eq 0 ] && echo " (no zones scanned yet)"
Fetch — Query Overpass API using the DaveHansenTiger user+timestamp filter with ["highway"] tag and out meta geom for full metadata. The query targets ways imported by DaveHansenTiger between 2007-08-03 and 2008-05-04 within the zone bbox. Retry logic: primary endpoint, 30s wait, primary again, then kumi mirror.
Classify — Assign defect classes:
highway=residential + oneway=yes + multi-segment name — compound defect, highest routing impactoneway=yes on residential streetsGap detection — Haversine endpoint analysis with 30m threshold + 5m junction clustering to find disconnected road segments
History filter (optional) — Analyze OSM revision history to determine if ways have been meaningfully reviewed since import. Skip with --skip-history for faster scans.
Reports — Generate XLSX workbook (8 sheets), interactive Leaflet dashboard, and 4 CSV slices
Start the web server if not running, then use the Scan tab at http://localhost:3000. Or run via Python:
from pathlib import Path
from osm.fetch import fetch_overpass
from osm.classify import classify
out_dir = Path(f"osm_audit_{zone_key}")
raw = fetch_overpass(zone_key, out_dir)
classified = classify(raw)
Diff OSM geometry and attributes against TIGER/Line 2024 to identify import drift, new roads, and name-field artifacts in MetroNow zones.
Generate a MapRoulette challenge from scan results for defect classes with high false-positive rates. Creates GeoJSON tasks constrained to MetroNow zone polygons.
Deep revision history analysis for specific OSM ways — fetches full version history from OSM API v0.6, identifies import bot vs human edits, classifies review status with confidence scores.
Submit corrections to OSM API v0.6 with full community compliance — proper changeset tags, size limits, rate-limit awareness, and dry-run support.
Decompress MetroNow docs so future-you (and fresh AI sessions) can pick the project back up cold without re-deriving everything. Use when the user says docs are confusing, plain, dense, hard to follow, missing context, jargon-heavy, jump straight to conclusions, or are hard to come back to after time away; asks why something is the way it is; says they can't tell what a term means; asks for a walkthrough, explainer, primer, or context refresher; says a section assumes too much; or wants visual aids. Diagnoses missing-middle gaps (undefined jargon, unstated WHY, skipped bridge steps, prose where a picture is needed) and rewrites sections using a fixed template — definition first, then bridge steps, then a load-bearing diagram, then code citations. Outputs to docs/explainers/<topic>.md or in-place edits. No VitePress, no static-site generator — just markdown that GitHub renders.
Unified code review and audit standards for the MetroNow Atlas TIGER Audit Console (https://github.com/AICincy/MetroNow.git). Use this skill when auditing any code in the MetroNow project, doing a PR review, checking code quality, validating accessibility, reviewing Docker config, or when someone asks "review this code," "audit the frontend," "check my PR," "what's wrong with this file," or "does this meet our standards." The codebase is vanilla JavaScript (IIFE pattern, no framework), single-file HTML with inline CSS/JS, IBM Plex typography, Leaflet maps, and an Express.js backend on port 3000 that shells out to the Python `osm` CLI. Covers JS, HTML, CSS, and Dockerfile review with severity-classified findings.