{"id":50666151,"url":"https://github.com/sbauwow/atlas-tx","last_synced_at":"2026-06-08T06:31:24.887Z","repository":{"id":356635626,"uuid":"1233415331","full_name":"sbauwow/atlas-tx","owner":"sbauwow","description":"Texas county intelligence for water, permits, public evidence, MCP, and field verification.","archived":false,"fork":false,"pushed_at":"2026-05-18T06:32:44.000Z","size":4289,"stargazers_count":1,"open_issues_count":1,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-05-18T06:43:34.929Z","etag":null,"topics":["agent-skill","civic-tech","mcp","nextjs","public-data","texas","water-quality"],"latest_commit_sha":null,"homepage":"https://atlastexas.org","language":"TypeScript","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/sbauwow.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":"AGENTS.md","dco":null,"cla":null}},"created_at":"2026-05-08T23:57:08.000Z","updated_at":"2026-05-18T05:18:14.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/sbauwow/atlas-tx","commit_stats":null,"previous_names":["sbauwow/atlas-tx"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/sbauwow/atlas-tx","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/sbauwow%2Fatlas-tx","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/sbauwow%2Fatlas-tx/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/sbauwow%2Fatlas-tx/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/sbauwow%2Fatlas-tx/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/sbauwow","download_url":"https://codeload.github.com/sbauwow/atlas-tx/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/sbauwow%2Fatlas-tx/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":34051768,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-26T15:22:16.424Z","status":"online","status_checked_at":"2026-06-08T02:00:07.615Z","response_time":111,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["agent-skill","civic-tech","mcp","nextjs","public-data","texas","water-quality"],"created_at":"2026-06-08T06:31:03.477Z","updated_at":"2026-06-08T06:31:24.881Z","avatar_url":"https://github.com/sbauwow.png","language":"TypeScript","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Atlas TX\n\nAtlas TX is a map-first intelligence system for Texas public-interest evidence.\n\nPublic launch surfaces:\n- Live site: https://atlastexas.org\n- GitHub repo: https://github.com/sbauwow/atlas-tx\n- Atlas TX MCP server: `packages/mcp-server/` via `npm run mcp:stdio`\n- Atlas TX agent skill: `skills/atlas-tx/SKILL.md`\n\nAtlas TX v1.0.0 is the public launch candidate aimed at judges, newsroom investigators, civic-tech analysts, and AI-agent workflows. The fastest way to understand the project is: open atlastexas.org, inspect the county maps, then use the MCP + skill surfaces to query the same evidence stack programmatically.\n\nIt is built to help people find, verify, and act on county-level signals across permits, water, hydrology, operators, governance, environmental burden, and field observations.\n\nThis is not another passive dashboard.\n\nAtlas TX is designed as an operating system for county evidence:\n- statewide county intelligence maps\n- public-record ingestion and normalization\n- API and MCP access to the same underlying evidence\n- watchlist and queue workflows for investigators and operators\n- Android-based field verification and water-testing missions\n\nThe product thesis is simple: Texas has too much fragmented public data, too much local variation, and too many decisions hidden inside county-by-county process. Atlas TX turns that fragmentation into a usable, governed evidence layer for humans, communities, and AI-native workflows.\n\n## Why Texas\n\nTexas is the right proving ground because it combines:\n- county-scale variation that actually matters\n- intense water, land-use, permitting, and infrastructure pressure\n- large public-data surface area spread across agencies and authorities\n- real need for field verification, not just desktop analysis\n\nAtlas is also shaped by an AI TX community framing.\n\nThat means the system is not built as a closed expert terminal for a tiny class of insiders. It is built in the spirit of a Texas AI community that wants better public tools, stronger evidence workflows, and more local capacity to investigate what is happening on the ground.\n\n## Why Atlas TX is compelling\n\nAtlas TX closes a gap that most public-data products leave open.\n\nThey show records. Atlas is built to help users investigate them.\n\nIt combines three layers that usually live in separate products:\n- county intelligence from public records\n- field-ready water and infrastructure verification workflows\n- agent-native analytical tooling for repeatable investigation\n\nThat creates something bigger than a website:\n- a county intelligence workspace for researchers, journalists, and local operators\n- a structured permit and water watch system\n- an Android field-missions product for evidence capture and review\n- a patent-pending system that links public-record intelligence, map workflows, and structured field verification\n\n## Why this wins over ordinary dashboards\n\nMost dashboards stop at display.\n\nAtlas TX is designed to go further:\n- start from county maps, not buried filters\n- connect records, operators, water signals, and field evidence in one workflow\n- expose the same evidence through UI, API, and MCP\n- support repeat investigation, not one-off browsing\n- preserve separation between authoritative records, modeled signals, and community submissions\n\nThe result should feel less like a static civic site and more like an evidence workstation for Texas.\n\n## Example workflow\n\n1. Start on the county map.\n   Spot a county with unusual permit pressure, water alerts, or mismatch signals.\n2. Launch a field mission.\n   Use the Android workflow to capture strip results, photos, notes, and site context in a constrained evidence flow.\n3. Review the evidence together.\n   Compare public records, modeled signals, and community submissions without collapsing them into one opaque score.\n\n## Public launch highlights\n\nAtlas TX is now being prepared as a public-facing release, not just a hackathon prototype. The launch story has four visible entry points:\n\n- `atlastexas.org` for judges and public users who need the live product first\n- county-map-first web workflows for exploration and drilldown\n- a local MCP server for agent-native access to the same evidence\n- repo-local Atlas TX skills that make the MCP surface usable from external agents\n\n### Why the MCP and skill matter\n\nAtlas TX is not only a website. It is also a reusable evidence system for AI-native workflows.\n\n- The MCP server exposes read-only, source-attributed Atlas TX tools over stdio and JSON dispatch.\n- The Atlas TX skill packages the right activation behavior, guardrails, and example invocation flow for external agents.\n- Judges can see the same product two ways: a live public site at `atlastexas.org` and an agent/tool path driven by MCP + skill.\n\n### Freshcrate positioning\n\nAtlas TX should be presented to Freshcrate as:\n- an open-source Texas public-interest intelligence system\n- a working MCP-enabled agent package / project\n- a repo with a live public destination (`atlastexas.org`) and a programmatic interface (`packages/mcp-server/`)\n\nThat means release messaging should explicitly call out the web product, the MCP surface, and the repo-local skill set together rather than treating them as separate side projects.\n\n## Hackathon judging criteria coverage\n\nThis submission is designed to score clearly against the four published judging criteria.\n\n### 1. Technical Execution \u0026 Completeness\nAtlas is not a mockup.\n\nThis repo already includes:\n- a working Next.js application with multiple production-style route surfaces\n- statewide county intelligence pages across analytics, water, permits, operators, watchlists, and maps\n- JSON API routes powering the same evidence model\n- a local MCP server for agent/tool access to the same system\n- structured telemetry, shared query-state parsing, refresh pipelines, and committed cache artifacts\n- an Android/mobile field-verification lane and citizen observation workflow\n- test, lint, and build validation paths\n\nIn other words: this is a real, multi-surface system, not a thin demo shell.\n\n### 2. Partner Ecosystem \u0026 Utility\nAtlas fits the AITX Community × Codex environment directly.\n\nIt is built around:\n- agent-native development and iteration during the Codex hackathon\n- MCP-compatible system access through `packages/mcp-server/`\n- Texas open-data utility, matching the event's Texas public-data track\n- a workflow that makes public records usable by both humans and AI agents\n\nThis is the area we most wanted to make explicit in the README: Atlas is not only a civic data app, it is a practical Codex-era agent workspace with an MCP surface and a clear public-data use case.\n\n### 3. Value \u0026 Impact\nAtlas targets a real problem with real users.\n\nThe value is not hypothetical:\n- investigators can find county hotspots faster\n- journalists can move from map to source evidence faster\n- communities can compare official records against field observations\n- operators can track permits, water signals, and county-level pressure in one place\n- future field teams can carry the workflow into Android missions and water-testing review\n\nThe core impact claim is simple: Atlas makes fragmented Texas public data materially more usable.\n\n### 4. Innovation \u0026 Execution\nAtlas pushes beyond a normal dashboard by combining:\n- county intelligence maps\n- public-record evidence normalization\n- agent/MCP access to the same evidence base\n- Android mission workflows\n- structured water-testing and field verification\n- explicit separation of authoritative, explanatory, and community evidence classes\n\nThat combination is the main innovation.\n\nThe system is trying to bridge desktop analysis, agent workflows, and real-world field verification in a single evidence architecture.\n\n## What Atlas TX is\n\nAtlas TX is not a generic consumer app and not a single-score black box.\n\nIt is a governed county-intelligence workspace that helps users:\n- start from a statewide county map\n- drill into water, permit, operator, and dependency signals\n- inspect contradictions and outliers in public data\n- open the underlying county, source, permit, and operator surfaces\n- use the same cached evidence through UI, API, and MCP tools\n\n## Current product surfaces\n\n### Human-facing routes\n- `/` — platform overview\n- `/analytics` — statewide analytics terminal with county map headliner, risk / pressure / oil views, movers, scatter, and watchlist-ready lanes\n- `/water` — county-map-first water explorer with operational risk, mismatch severity, and TXG31 oil-and-gas extraction footprint\n- `/water/network` — county dependency and hydrology relationship workspace\n- `/water/counties/[slug]` — county water detail\n- `/water/sources/[slug]` — source provenance and anomaly detail\n- `/counties` — county explorer\n- `/counties/[slug]` — county intelligence page\n- `/permits` — permit pressure workspace\n- `/permits/[tceqId]` — permit detail\n- `/operators` — operator index\n- `/operators/[slug]` — operator detail\n- `/watchlists` — saved watchlists\n- `/data` — dataset registry\n- `/data/botnet` — ingest botnet operator view\n- `/map` — interactive map client\n- `/citizen` — isolated citizen observation prototype lane\n- `/glossary` — term definitions\n- `/education` — supporting explainers\n\n### API routes\n- `/api/health`\n- `/api/beacon` — first-party pageview/click pixel telemetry\n- `/api/event` — structured telemetry event ingest\n- `/api/counties/overview`\n- `/api/counties/[slug]`\n- `/api/ops/botnet`\n- `/api/permits/locations`\n- `/api/tiles/[z]/[x]/[y]`\n- `/api/watchlists`, `/api/watchlists/*`\n- `/api/water/overview`\n- `/api/water/counties/[slug]`\n- `/api/water/oil-gas-extraction`\n- `/api/water/alerts`\n- `/api/water/gauges`\n- `/api/water/sources/[slug]`\n- `/api/water/sources/network`\n- `/api/water/sources/network/hydrology`\n- `/api/water/fema/nfhl/*`\n- `/api/water/lcra/*`\n- `/api/water/gbra/*`\n- `/api/citizen/observations`, `/api/citizen/observations/[id]`\n- `/api/scrape/plan`\n\n## Core workflows\n\n### 1. Map-first county analysis\nAtlas leads with county maps, not tables.\n\n- `/analytics` emphasizes statewide county risk, permit pressure, and TXG31 oil extraction density.\n- `/water` emphasizes operational water risk, mismatch severity, and oil-and-gas extraction permit footprint.\n- Supporting tables, movers, and scatterplots are secondary validation layers.\n\n### 2. County-to-detail drilldown\nA user should be able to move from statewide map → county selection → county detail → source / permit / operator evidence without changing mental model.\n\n### 3. Evidence parity across UI, API, and MCP\nThe same cached and normalized public-data layers are intended to power:\n- the Next.js UI\n- JSON API routes\n- local MCP tools\n- scripted refresh / pipeline health utilities\n\n### 4. Watchlist-ready analytical handoff\nAtlas includes watchlist and queue-style surfaces so a user or agent can save, rank, and reopen counties, operators, and analytical leads.\n\n## Data lanes\n\n### Texas core sources\n- `7fq8-wig2` — TCEQ Water Quality Individual Permits (Active/Pending)\n- `6pm5-am5m` — TCEQ General Water Permits\n- `hr84-s96f` — Texas Water Districts\n- additional TCEQ/TWDB/authority lanes as documented in the repo contracts and wiki\n\n### Federal joins and national context\n- EPA SDWIS\n- EPA EJScreen\n- Census ACS 5-year\n- FEMA NFHL\n- USGS NWIS\n- NOAA/NWS\n\n### Regional authority and basin context\nAtlas also includes basin-specific and authority-specific lanes where useful, including LCRA and GBRA integrations, with explicit caveats when a lane is basin-scoped rather than statewide.\n\n## Derived / normalized signals\n\nCurrent and planned signal families include:\n- county risk signals\n- permit pressure signals\n- TXG31 oil-and-gas extraction permit counts\n- TXG34 petroleum bulk permit counts\n- county mismatch signals\n- source dependency signals\n- downstream hydrology dependency scores\n- governance density\n- floodplain coverage proxies\n- stream gauge and alert context\n- county-month research artifacts for water-risk modeling\n\nAtlas treats these as evidence layers, not final causal verdicts.\n\n## Architecture\n\n### Compact architecture diagram\n\n```text\n                        public/open datasets\n   TCEQ / TWDB / EPA / FEMA / USGS / NOAA / regional authorities\n                                  |\n                                  v\n                 src/lib/* fetchers + normalizers + scoring services\n                                  |\n                                  v\n      public/cache/* \u003c---- scripts/refresh:* ----\u003e pipeline-health.json\n                                  |\n                    +-------------+-------------+\n                    |                           |\n                    v                           v\n           Next.js route pages            JSON API routes\n   /analytics /water /permits /counties   /api/water/* /api/counties/*\n                    |                           |\n                    +-------------+-------------+\n                                  |\n                                  v\n                     map-first county intelligence UX\n                                  |\n                    +-------------+-------------+\n                    |                           |\n                    v                           v\n              watchlists + telemetry       local MCP server\n            /watchlists /api/beacon        packages/mcp-server/\n                 /api/event\n```\n\n### App/runtime\n- Next.js 16 App Router\n- React 19\n- TypeScript\n- Tailwind\n- Prisma\n- SQLite for local app/stateful prototype lanes\n\n### Major directories\n- `src/app/` — Next.js routes, pages, API endpoints\n- `src/lib/` — normalization, fetchers, scoring, telemetry, query helpers, county logic, water services\n- `src/app/components/` — reusable app UI and chart shells\n- `packages/mcp-server/` — local MCP surface\n- `scripts/` — refresh pipelines and dataset builders\n- `public/cache/` — committed or generated cache artifacts used by the app\n- `docs/` — contracts, plans, design, wiki, state, ownership, research\n- `skills/` — repo-local agent skills\n- `tests/` — Vitest coverage\n\n### Key route-to-file map\n\n| Surface | Primary file | Notes |\n|---|---|---|\n| `/` | `src/app/page.tsx` | product overview |\n| `/analytics` | `src/app/analytics/page.tsx` | statewide map-first analytics terminal |\n| `/water` | `src/app/water/page.tsx` | county-map-first water workspace |\n| `/water/network` | `src/app/water/network/page.tsx` | dependency + hydrology view |\n| `/water/counties/[slug]` | `src/app/water/counties/[slug]/page.tsx` | county water detail |\n| `/water/sources/[slug]` | `src/app/water/sources/[slug]/page.tsx` | source anomaly/provenance |\n| `/counties` | `src/app/counties/page.tsx` | county index |\n| `/counties/[slug]` | `src/app/counties/[slug]/page.tsx` | county intelligence detail |\n| `/permits` | `src/app/permits/page.tsx` | permit pressure workspace |\n| `/permits/[tceqId]` | `src/app/permits/[tceqId]/page.tsx` | permit detail |\n| `/operators` | `src/app/operators/page.tsx` | operator index |\n| `/operators/[slug]` | `src/app/operators/[slug]/page.tsx` | operator detail |\n| `/watchlists` | `src/app/watchlists/page.tsx` | saved watchlists |\n| `/data` | `src/app/data/page.tsx` | dataset registry |\n| `/data/botnet` | `src/app/data/botnet/page.tsx` | ingest botnet operator view |\n| `/map` | `src/app/map/page.tsx` | map shell entry |\n| `/api/event` | `src/app/api/event/route.ts` | structured telemetry ingest |\n| `/api/beacon` | `src/app/api/beacon/route.ts` | pixel telemetry ingest |\n| `/api/ops/botnet` | `src/app/api/ops/botnet/route.ts` | botnet health + queue summary |\n| `/api/water/overview` | `src/app/api/water/overview/route.ts` | statewide water summary |\n| `/api/water/oil-gas-extraction` | `src/app/api/water/oil-gas-extraction/route.ts` | TXG31 permit lane |\n| local MCP server | `packages/mcp-server/src/index.js` | command entrypoint |\n\n### Important internal modules\n- `src/lib/water/water-summary-service.ts` — water overview + county breakdown orchestration\n- `src/lib/water/tceq-general-permits.ts` — permit normalization and lane classification (`TXG31`, `TXG34`, residual)\n- `src/lib/query-params.ts` — shared parsing for route-state parameters\n- `src/lib/telemetry/*` — local telemetry envelope and client helpers\n- `src/app/components/track.ts` — pixel-style beacon tracking\n- `src/app/components/tracked-link.tsx` — structured event tracking on link interactions\n- `src/app/components/page-view-beacon.tsx` — pageview beacon client hook\n\n## Telemetry model\n\nAtlas now ships two first-party telemetry paths:\n\n### 1. Beacon path\n- route: `/api/beacon`\n- purpose: low-friction pageview/click-style tracking via image pixel\n- current usage: pageview and lightweight link/event emission\n\n### 2. Structured event path\n- route: `/api/event`\n- purpose: normalized event ingestion with session ID and envelope metadata\n- envelope fields:\n  - `app`\n  - `env`\n  - `release`\n  - `user_id`\n  - `session_id`\n  - `event_name`\n  - `props`\n  - `ts`\n\nThis split preserves fail-open beacon tracking while enabling richer product telemetry across map mode, county selection, watchlists, and drilldowns.\n\n## Query-state model\n\nAtlas uses URL-driven analytical state heavily.\n\nShared helpers now live in:\n- `src/lib/query-params.ts`\n\nCurrent helper set:\n- `getFirstQueryParam`\n- `parseEnumQueryParam`\n- `resolveAllowedQueryParam`\n- `clampQueryText`\n- `parsePositiveIntQueryParam`\n\nThese are used to stabilize mode/county parsing across analytics and water surfaces and are intended to expand to permits, operators, and map routes.\n\n## MCP server\n\nAtlas ships a local MCP surface in `packages/mcp-server/`.\n\nRun it with:\n\n```bash\nnpm run mcp -- discover_datasets\nnpm run mcp -- list_permit_filing_red_flags '{\"county\":\"Travis County\",\"limit\":5}'\nnpm run mcp -- get_permit_filing_detail '{\"tceq_id\":\"WQ0000447000\"}'\nnpm run mcp -- build_permit_protest_prep '{\"tceq_id\":\"WQ0000447000\"}'\nnpm run mcp -- list_county_pending_fights '{\"county\":\"Travis County\",\"limit\":5}'\nnpm run mcp -- get_pipeline_health\nnpm run mcp -- get_roadmap_open_data_queue\n```\n\nSee:\n- `docs/contracts/mcp-tools.md`\n- `docs/contracts/skill-protocol.md`\n\n## Refresh and pipeline operations\n\n### Local development\n```bash\nnpm install\nnpm run dev\n```\n\n### Build/test/lint\n```bash\nnpm test\nnpm run lint\nnpm run build\n```\n\n### Data refresh entrypoints\n```bash\nnpm run refresh:all\nnpm run refresh:botnet\nnpm run refresh:weather\nnpm run refresh:roadmap-open-data\nnpm run refresh:cid\nnpm run refresh:surface-water-quality\nnpm run refresh:city-open-data\nnpm run refresh:city-open-data-curated\nnpm run refresh:city-open-data-ranked\nnpm run refresh:twdb-hydrology\nnpm run refresh:county-month-precipitation\nnpm run refresh:county-month-streamflow\nnpm run refresh:county-month-drought\nnpm run refresh:county-month-temperature\nnpm run refresh:county-month-nws-flood-alerts\n```\n\n### Agentic ingest botnet\n\nAtlas now includes a registry-driven ingest spine for weather and future roadmap datasets.\n\nKey pieces:\n- `config/execution-registry.county.json` — machine-readable county ingest roadmap\n- `src/lib/execution/execution-registry.ts` — typed loader/helpers for the execution registry\n- `src/lib/atlas-ingest-orchestrator.ts` — dependency-aware task catalog and botnet runner\n- `scripts/refresh-weather.ts` — grouped county-month weather refreshes\n- `scripts/refresh-roadmap-open-data.ts` — future open-data candidate queue snapshot\n- `scripts/refresh-all.ts` — top-level orchestrator that stages core water, weather, catalog, roadmap, then CID work\n\nThe orchestrator is designed around waves:\n- wave 0 — current county water backbone\n- wave 1 — county-month weather and hydrology history\n- wave 2 — SDWIS / EJ / ACS / ECHO / CID deepening\n- wave 3+ — boil-water notices, E2, IBI, and later community-verification lanes\n\n### Pipeline outputs\n- `public/cache/pipeline-health.json` — refresh pipeline health/status\n- `public/cache/*` — analytics and dataset artifacts consumed by pages and APIs\n\n## Technical documentation index\n\n### Primary operational docs\n- `AGENTS.md` — development guide for agents and contributors\n- `docs/STATE.md` — current operational/project state\n- `docs/OWNERSHIP.md` — path and workstream ownership\n- `docs/plans/README.md` — plan index\n\n### Contracts\n- `docs/contracts/dataset-registry.md`\n- `docs/contracts/execution-registry.md`\n- `docs/contracts/mcp-tools.md`\n- `docs/contracts/skill-protocol.md`\n- `docs/contracts/community-observation.md`\n- `docs/contracts/county-month-water-risk-panel.md`\n\n### Design and execution plans\n- `docs/design/atlas-missions-design-memo.md`\n- `docs/design/agent-build-workflow.md`\n- `docs/plans/2026-05-09-atlas-of-maps-reframe.md`\n- `docs/plans/2026-05-09-mcp-and-pipeline-automation.md`\n- `docs/plans/2026-05-10-county-dataset-roadmap.md`\n- additional dated plans under `docs/plans/`\n\n### Research and wiki\n- `docs/research/` — method and source research\n- `docs/wiki/index.md` — wiki index\n- `docs/wiki/datasets/` — per-dataset notes\n- `docs/wiki/concepts/` — domain vocabulary\n- `docs/wiki/agencies/` — agency references\n- `docs/wiki/comparisons/` — source comparisons\n- `docs/wiki/episodes/` — execution log slices\n\n### Repo-local skills\n- `skills/atlas-tx-bootstrap/SKILL.md`\n- `skills/atlas-tx-public-data-lanes/SKILL.md`\n- `skills/atlas-tx-water-data-lane/SKILL.md`\n- `skills/atlas-tx-mcp-pipeline-automation/SKILL.md`\n- `skills/atlas-tx-pending-permits-dashboard/SKILL.md`\n- `skills/atlas-tx-permit-red-flags-protest-helper/SKILL.md`\n- `skills/atlas-tx-regional-water-authority-ingestion/SKILL.md`\n- `skills/atlas-tx-glossary-tooltips/SKILL.md`\n- `skills/atlas-tx-web-style-taxonomy/SKILL.md`\n- `skills/atlas-tx/SKILL.md`\n\n## Documentation completeness checklist\n\nThis README is intended to cover the platform at the level of a public technical overview.\n\nCovered here:\n- product purpose\n- route surfaces\n- API surfaces\n- data lanes\n- architecture\n- telemetry model\n- query-state model\n- MCP surface\n- refresh/pipeline commands\n- documentation index\n- testing/build workflow\n- guardrails and constraints\n\nCanonical deeper docs live in `docs/`, `skills/`, and `AGENTS.md`.\n\n## Citizen observation and Android field lane\n\nAtlas includes a separate, non-regulatory citizen observation prototype at `/citizen`, and the broader system roadmap includes a mission-driven Android application for field verification.\n\nThe Android/mobile direction is not a generic upload app. It is intended to support:\n- mission-based field collection\n- guided water-strip capture\n- site-context notes and corroborating photos\n- repeat monitoring loops\n- confidence-aware evidence submission\n- later integration with map, county, and source workflows\n\nThe current mobile/field product direction is documented in:\n- `docs/design/android-missions-mobile-flow.md`\n- `docs/contracts/community-observation.md`\n- `docs/research/smartphone-colorimetry.md`\n\n### Inference provider\n\nThe Android client never calls an inference API directly and never holds an\ninference key. Strip photos are uploaded to `/api/citizen/observations`, and\nthe route runs the server-side vision pass against\n[Featherless](https://featherless.ai) (preferred, OpenAI-compatible chat\ncompletions at `https://api.featherless.ai/v1`) when `FEATHERLESS_API_KEY`\nis set. OpenAI `gpt-4o-mini` is the fallback when only `OPENAI_API_KEY` is\nset; if neither is set the observation is recorded with the client reading\nonly and routed to manual review. Default Featherless model is\n`meta-llama/Llama-4-Scout-17B-16E-Instruct`; override with\n`FEATHERLESS_MODEL`. The actual model that ran is recorded on each\nobservation as `llmModel`. Implementation: `src/lib/observations/vision.ts`.\n\n### Water testing and field verification\n\nA major Atlas system lane is structured water testing and field verification.\n\nThat lane is designed around:\n- smartphone-assisted strip/colorimetry workflows\n- constrained capture rather than arbitrary photo uploads\n- confidence scoring and QA flags\n- explicit separation between community screening and regulatory/public-record evidence\n- county/watershed aggregation as additive context rather than silent score modification\n\nImportant constraints:\n- it is isolated from the main scoring stack\n- it does not feed county risk scoring or water mismatch scoring by default\n- it is a prototype workflow for strip-photo observation capture and review\n- it should not be interpreted as a compliance, diagnostic, or regulatory measurement system\n- it is best framed as screening, verification, and lead-generation evidence\n\n## Patent-pending system framing\n\nAtlas TX is being developed as part of a broader patent-pending system concept spanning:\n- map-first county intelligence\n- public-record permit and water evidence integration\n- mobile mission workflows\n- structured community water-testing capture\n- confidence-scored field verification\n- explicit separation of authoritative, explanatory, and community evidence classes\n\n### Patent-pending system components\n\n```text\ncounty intelligence maps\n        +\npublic-record permit + water evidence\n        +\nAndroid mission workflows\n        +\ncommunity water-testing capture\n        +\nconfidence-aware field verification\n        +\nevidence-class separation and review\n```\n\nThis repository is the open technical platform layer for that system. It should describe the product honestly and compellingly, while avoiding unsupported legal, medical, or regulatory claims.\n\n## Development workflow\n\nAtlas has been built through an agent-assisted workflow centered on Hermes Agent as the orchestration layer, with repo-local skills, MCP tooling, and iterative map/API/test slices inside this repository.\n\nPublic README copy intentionally stays focused on the platform and its technical surfaces rather than on specific model-vendor branding.\n\nFor the internal workflow summary, see:\n- `docs/design/agent-build-workflow.md`\n\n## Constraints and product guardrails\n\n- public/open data only, with source attribution\n- no medical or diagnostic claims\n- no unsupported causal claims\n- no investor-grade or ratings-agency claims\n- burden/exposure indicators are described as indicators, not proof of harm\n- every serious score or summary should remain explainable by source datasets, joins, and caveats\n- basin-specific lanes must stay clearly labeled as basin-specific, not statewide truth\n\n## Contributing\n\nStart with:\n- `AGENTS.md`\n- `docs/OWNERSHIP.md`\n- `docs/STATE.md`\n\nThen run:\n```bash\nnpm install\nnpm test\nnpm run lint\nnpm run build\n```\n\nIf you are touching route-state, telemetry, map modes, or water summary logic, update the relevant tests in `tests/` in the same slice.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fsbauwow%2Fatlas-tx","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fsbauwow%2Fatlas-tx","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fsbauwow%2Fatlas-tx/lists"}