outdooriq-mcp
Provides lake conditions, fish-stocking records, fishing-favorability scoring, and live weather for outdoor recreation AI agents, powering trip-planning and fishing apps with 72,000+ US lakes and 293,000+ stocking events.
README
OutdoorIQ MCP — Outdoor Recreation Intelligence
Powered by 72,000+ US lakes and 293,000+ stocking events across 12 states.
OutdoorIQ MCP is a paid MCP server that exposes lake conditions, fish-stocking records, fishing-favorability scoring, and live weather to AI agents. It is built on the CastIQ dataset (the same data that powers fishing-seo.pages.dev and the CastIQ catalog APIs).
If you're building a trip-planning assistant, a fishing app, a travel concierge, or an outdoor-brand agent that needs to recommend lakes, time visits to recent stocking events, or pull a fishing report on demand, this is the data layer you wire up.
For consumer-facing AI products, OutdoorIQ replaces a ten-source ETL with one authenticated MCP endpoint — and bills predictably so you don't get a surprise S3 invoice the first weekend traffic spikes.
Pricing
| Tier | Price | Limits |
|---|---|---|
| Free | $0 | outdooriq-dev-key-001, 50 calls/day |
| Pro | $14/mo | Unlimited |
| Pay-as-you-go | $0.01/call | No monthly minimum |
Listing & checkout: https://mcpize.com/outdooriq-mcp
Install in Claude
Live URL: https://mcp.castiq.net/mcp (Railway fallback: https://web-production-9b8950.up.railway.app/mcp)
The fastest path uses mcp-remote as a stdio→HTTP bridge:
claude mcp add outdooriq-mcp -- npx -y mcp-remote \
https://mcp.castiq.net/mcp \
--header "X-API-Key:outdooriq-dev-key-001"
Or configure manually:
{
"mcpServers": {
"outdooriq-mcp": {
"url": "https://mcp.castiq.net/mcp",
"headers": { "X-API-Key": "outdooriq-dev-key-001" }
}
}
}
Anthropic MCP Registry entry: io.github.bch1212/outdooriq-mcp —
listed at https://registry.modelcontextprotocol.io.
Example agent prompt:
"Find top 5 trout lakes near Chicago for this weekend."
The agent calls get_nearby_lakes (lat/lng of Chicago, radius 200mi),
filters for species: trout via get_top_lakes, then pulls
get_fishing_report_summary for the top match.
Tool Reference
| Tool | Args | Returns |
|---|---|---|
search_lakes |
name?, state?, county?, min_acres?, max_acres?, limit? |
List of matching lakes |
get_lake_details |
lake_id |
Coords, acreage, depth, species, facilities |
get_stocking_data |
lake_id?, species?, year?, limit? |
Recent stocking events |
get_fishing_score |
lake_id |
0-100 score with bucket breakdown |
get_nearby_lakes |
lat, lng, radius_miles?, min_score?, limit? |
Lakes near GPS, sorted by score |
get_weather_for_lake |
lake_id |
Current + 7-day forecast (Open-Meteo) |
get_top_lakes |
state?, species?, limit? |
Highest-scoring lakes |
get_stocking_schedule |
state?, species?, month?, limit? |
Most-recent matching events as a planning prior |
search_species |
state?, season? |
Actively-stocked species |
get_fishing_report_summary |
lake_id |
Natural-language report |
Architecture
client (Claude / agent)
│ HTTP POST /mcp (JSON-RPC 2.0)
▼
FastAPI app (server.py)
│ X-API-Key auth + per-day rate limiter
▼
Tool registry (10 tools)
│
▼
db.connection.py ──────┐
│ Postgres mode │ → CastIQ Postgres (72k lakes, 293k stockings)
│ SQLite fallback│ → bundled seed (100+ lakes, ~150 stockings)
▼
tools.* (lakes, stocking, scoring, weather, reports)
│
▼
Open-Meteo (no key)
Postgres vs SQLite mode
The server picks a backend at startup:
- If
DATABASE_URLis set, it triesasyncpg.create_pool. On success → logs[OutdoorIQ] Running in Postgres mode. - If the pool fails (timeout, bad creds, DB down) or
DATABASE_URLis unset, the server seeds an in-memory SQLite DB with 100+ real lakes and logs[OutdoorIQ] Running in SQLite fallback mode.
The server always starts, regardless of Postgres availability.
| Capability | Postgres mode | SQLite fallback |
|---|---|---|
| Lake catalog | 72,669 (12 states) | ~106 (WI, MN, IL, IA, MO) |
| Stocking events | 293,821 historical | ~150 templated, recency-tuned |
| Species coverage | All states/species in CastIQ | Curated subset (walleye, bass, musky, trout, crappie, perch, pike, panfish, salmon, lake_trout, sauger, white_bass, sturgeon, catfish, bluegill, smallmouth_bass) |
| Year-over-year analysis | Yes — multi-year stockings | Limited — events are anchored relative to "now" |
| Stocking-schedule tool | Real historical patterns | Approximated from seed |
| Weather, scoring, reports | Identical | Identical |
Local development
git clone <this repo>
cd mcp-outdoors
python3 -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
# Run with SQLite fallback (no DATABASE_URL needed)
python -m run
# OR run against the CastIQ Postgres
export DATABASE_URL=postgresql://vikinetic:vikinetic_dev@localhost:5444/vikinetic
python -m run
Then:
curl -s http://localhost:8080/health
curl -s -X POST http://localhost:8080/mcp \
-H "X-API-Key: outdooriq-dev-key-001" \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'
Tests
pytest -v
The test suite covers both the SQLite fallback path and the Postgres dispatch
path (via a mocked asyncpg pool), plus auth, rate limits, all 10 tools,
JSON-RPC initialize / list / call, and the scoring algorithm's bucket math.
Deploy to Railway
A deploy.sh script is included at the repo root. Run it on your Mac (the
Cowork sandbox can't reach Railway/Stripe/Cloudflare APIs):
./deploy.sh
It expects RAILWAY_API_TOKEN (or RAILWAY_TOKEN exported as RAILWAY_API_TOKEN),
points the project at nixpacks.toml, and sets the DATABASE_URL env var if
you've also provisioned the CastIQ Postgres on Railway.
Railway gotcha (already handled): Railway exec's
startCommandwithout a shell, so$PORTdoesn't expand. We usepython -m runand readPORTfromos.environinsiderun.py.
MCPize listing copy
OutdoorIQ MCP — the data layer for outdoor-rec AI agents. 72,000+ US lakes, 293,000+ fish-stocking events, and live weather behind one authenticated endpoint. Search lakes by name, state, or acreage; pull stocking history filtered by species and month; compute a 0-100 fishing-favorability score that bakes in recency, species diversity, lake size, and current conditions. One JSON-RPC call replaces a multi-source ETL.
Built for fishing apps, trip-planning assistants, travel concierges, and outdoor-brand agents. $14/mo Pro for unlimited use, or pay $0.01 per call. Free dev tier (50 calls/day) lets you ship a prototype before opening your wallet. Install with
claude mcp add outdooriq-mcp --url https://mcp-outdoors.up.railway.app/mcp.
License
MIT. See LICENSE.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.
E2B
Using MCP to run code via e2b.