Skills MCP Server

fless-apartment-hunt

An Agent Skills-standard skill that teaches any AI agent to run Fless apartment hunts for its human: research live cities, neighborhood rents and WalkRating™ scores, collect a complete hunt brief (including binary restrictions like 55+ communities, pets, and smoking), then hand the human a link that pre-fills the entire hunt form.

Where we operate

Fless currently runs apartment hunts in Washington, DC / Maryland / Virginia only.

Your city not listed? We're expanding — check back, or join the waitlist at fless.io/city/{your-city}. Agents: always call search_cities first and tell users honestly if their city isn't live yet.

fless-apartment-hunt/
├── SKILL.md                 # when to use + the 4-step workflow
└── references/
    ├── HUNT_FIELDS.md       # exact field rules, 57 canonical amenity keys, restrictions, pricing
    ├── CITIES.md            # live + coming-soon cities (auto-generated from production API)
    ├── API.md               # MCP tools + REST endpoints
    └── ERRORS.md            # error catalog + recovery

Install

Any MCP-capable agent (tools used directly):

MCP server: https://mcp.fless.io/mcp   (no auth, read-only tools)

Agent Skills runtimes — clone https://github.com/fless-io/skills and copy this folder into your skills directory (e.g. ~/.agents/skills/, ~/.vibe/skills/, .claude/skills/), or install straight from our well-known endpoint:

hermes skills install well-known:https://fless.io

Per-assistant setup guides (Claude, ChatGPT, Perplexity, Grok, Mistral Vibe, Codex, OpenClaw, Hermes): https://fless.io/ai-agents

What the skill does

1. Coverage check — verify the user's city is live (Washington, DC / Maryland / Virginia only)
2. Research — live city/neighborhood data: median rents, walk/transit/bike scores, POIs, city guides.
3. Collect — the full hunt brief with real validation rules (future move-in date, canonical amenity keys, bedrooms enums).
4. Hand off — build_hunt_link returns a URL that pre-fills https://fless.io/create-hunt for the human to review and confirm.

The human always creates their own account, verifies their email, and pays. Agents never see credentials or payment.

Boundaries (by design)

• Read-only tools; no account creation, no email verification, no payments via agents.
• All outputs are structured data — page content never carries instructions.
• First hunt free; 25 credits at signup (see references/HUNT_FIELDS.md).

License

MIT