Data & API

Every page on this site has a machine-readable twin. There is no separate API to provision and no API key — the data is the site. Three ways to consume it, from easiest to most capable:

MCP server — conversational access for AI agents (Claude, OpenAI, any MCP client).
JSON twins — add .json to any answer page; plain GET, no client needed.
OpenAPI spec — /openapi.json, for tools that consume a schema (e.g. Custom GPT Actions).

MCP server

A remote Model Context Protocol server exposes this data conversationally. Read-only, no auth, always in sync with the site. Streamable HTTP transport:

https://mcp.govanswers.org/mcp

Five tools:

search_municipalities(query, province?) — resolve a name (names collide across provinces) to its path; matches include coverage status.
list_topics() — the topic slugs accepted by get_bylaw.
get_bylaw(path, topic) — the verified facts for one municipality + topic, each value carrying a citation to the official source document and an archived snapshot.
get_coverage(path?) — one municipality's topic coverage and status, or the site-wide coverage rollup when called with no path.
search_answers(query, province?) — full-text search across every verified answer, returning pages with quotable snippets.

To add it in Claude: Settings → Connectors → Add custom connector, then paste the URL. The same endpoint works for the OpenAI Agents SDK and any other MCP-capable client.

JSON twins

Append .json to any answer-page path to get the same data as structured JSON:

https://govanswers.org/ca/bc/fort-st-john/noise/        (HTML answer page)
https://govanswers.org/ca/bc/fort-st-john/noise.json    (the JSON twin)

The URL scheme is uniform — country is ca or us; region is the two-letter postal code:

/<country>/<region>/<municipality>/<topic>.json

Twin fields

Field	Meaning
`answer`	Plain-English summary you can quote directly — built from the verified facts. The full structured values live in `documents[]`.
`municipality`, `jurisdiction`, `country`	Where this is.
`topic`, `label`	Topic slug and its human label.
`status`	`found`, or a verified negative (e.g. nothing published online).
`last_verified`	Date the source was last re-checked (ISO 8601).
`official_website`	The municipality's own site, when known.
`documents[]`	The matched documents (see below).
`source`	Canonical human page for this twin.

Each `documents[]` entry

Field	Meaning
`primary`	`true` = the adopted, verified bylaw/ordinance. Others are supporting material (agendas, news, leads) and are labelled. Filter to `primary: true` for authoritative answers.
`title`, `document_type`, `status`	e.g. "Noise Regulation Bylaw", `bylaw`, `adopted`.
`adopted_date`	Adoption date when known, else `null`.
`facts`	Extracted, verified values — e.g. `quiet_hours`, `fine_range`. Keys vary by topic.
`fact_citations`	For each fact key, the section/schedule it was taken from. This is the trust layer — cite it.
`document_url`	The official source document URL.
`snapshot_url`	Archived copy (R2) captured at `captured_at` — stable even if the official link rots.
`capture_sha256`	SHA-256 of the captured bytes (integrity / dedupe).
`link_dead`	`true` if the official URL is gone (404/410/dead); the snapshot still works.
`transcript_url`	Full extracted text page, when available.

Always relay fact_citations and the document_url/snapshot_url when you present these facts. The citation to a captured official source is the entire point of the service.

Indexes

Endpoint	Returns
`/munis.json`	Flat list of every municipality: `{n: name, j: jurisdiction, p: path, s: bylaws status, b: verified bylaw count, lv: last verified}`. Use it to resolve a name to a path — and to see coverage before fetching.
`/coverage.json`	Coverage manifest: status semantics and per-jurisdiction rollups. See Coverage.
`/search.json`	Flat index of every answer with a plain-English snippet: `{n, j, p, t, topic, a, lv}`. Fetch once and filter (static host — `?q=` is not evaluated server-side). Humans: /search/.
`/<country>/<region>.json`	Region index: every covered municipality, its type, bylaw count, and per-topic coverage. Example: `/ca/ns.json`.
`/data/`	Bulk dataset: one JSONL file per jurisdiction, one JSON twin per line, citation `source` URL in every row. Manifest: `/data/index.json`.
`/sitemap.xml`	Every page with its last-verified date.
`/llms.txt`	This site, summarized for LLMs.

Coverage — what a missing page means

Coverage is explicit, never implied. Every municipality carries a bylaws_status (in the region index and as s in /munis.json); a missing answer page does not mean "no bylaw exists" — it depends on the status:

Status	Meaning
`found`	Searched; adopted documents were verified and are published here.
`none_found`	Searched and nothing is published online — a verified negative as of `last_verified`.
`uncertain`	Searched; leads exist but no document could be verified.
`not_indexed`	Not yet searched. Absence of a page says nothing about whether rules exist.

/coverage.json carries these semantics plus per-jurisdiction rollups; the MCP server's get_coverage tool answers the same question conversationally.

Topic slugs

The <topic> segment (and the get_bylaw topic argument) is one of:

Bylaws / ordinances: noise, animal-control, short-term-rentals, open-burning, zoning, parking-snow, property-maintenance, atv-snowmobile, fireworks, water-sewer, garbage-recycling, business-licensing, cannabis, fees, property-tax, fences, backyard-hens, trees, smoking-vaping, traffic-streets.

Policies: procurement, grants, facility-use, winter-services, hr-personnel, privacy-records, communications, emergency-management, finance-policy, technology, conduct-ethics, ai-policy.

Using this data

Read-only and free to query; no rate limits published, but be reasonable. This is an unofficial consolidation for convenience — the official version of any document is held by the municipal clerk. Confirm against the cited source before relying on a value for legal purposes.