The site is written for people and for machines
Every page will have a Markdown twin, every content type an API, and an MCP endpoint will give tools the same access with the same rules as everyone. This page describes what is coming; none of it is live yet.
api.paxabyssi.com/api/v1/
| Route | Returns |
|---|---|
| GET /api/v1/me | The signed-in account and its permissions |
| GET /api/v1/wiki/pages/{title} | A wiki page: stable and latest revision, Pax Markdown source, infobox JSON |
| POST /api/v1/wiki/pages/{title}/revisions | A new revision; edits from new accounts land in pending changes |
| GET /api/v1/articles | Features, devlogs and news, cursor paginated |
| GET /api/v1/forum/c/{community}/posts | A community's feed with its sort |
| GET /api/v1/sim/stars/{hyg_id} | A star's catalogue row with its corrections and provenance |
| GET /api/v1/sim/systems/{stem} | A generated system: bodies, elements, epoch, belts, stations |
Conventions
- JSON, with one error envelope:
{error: {code, message, details}} - Cursor pagination:
{results, next_cursor} - Browsers use the session cookie and a CSRF header; scripts use
Authorization: Bearer pax_pat_... - Personal access tokens are hashed at rest and scoped; bot tokens are separate and visibly labelled
- Every permission check goes through one policy; the API and the MCP server call the same service functions
mcp.paxabyssi.com/mcp
A Model Context Protocol endpoint over the same services as the API: 30 tools, reference resources and prompt templates. Reading needs no token. Writing takes a scoped token and follows the same permissions, review and rate limits as the website, and every call is logged.
Connect a client
{
"mcpServers": {
"paxabyssi": {
"type": "http",
"url": "https://mcp.paxabyssi.com/mcp",
"headers": {
"Authorization": "Bearer ${PAX_MCP_TOKEN:-}"
}
}
}
}- Clients configured by file: Add the block above to the client's mcp.json. The token comes from an environment variable and never sits in the file.
- Clients with a settings form: Endpoint https://mcp.paxabyssi.com/mcp, transport Streamable HTTP, header Authorization: Bearer <token>.
- Clients without custom headers: Connect with no authentication to read. Signing in with OAuth for writes comes in a later phase.
- Tokens: Create one at /settings/tokens and choose its scopes. Revoke it there at any time.
| Area | Tools and what they need |
|---|---|
| Site | site_infoWho you are connected as, what your token may do, the house style |
| Wiki, read | wiki_search · wiki_get_page · wiki_get_history · wiki_diff · wiki_list_changes · wiki_list_category · wiki_linksNo token |
| Wiki, write | wiki_create_page · wiki_edit_page · wiki_talk_post · wiki_move_page · wiki_revert · wiki_reviewwiki:edit, wiki:create, wiki:move, wiki:review |
| Media | media_search · media_get · media_upload · media_update_metadataReads free; uploads need media:upload and full provenance |
| Articles | article_list · article_get · article_create_draft · article_update_draft · article_submit_for_review · article_publishReads free; drafting and publishing are staff only |
| Forum | forum_list · forum_get_thread · forum_search · forum_create_thread · forum_reply · report_contentReads free; posting is staff and TL2 at launch |
What you may take, and on what terms
The star catalogue is derived from the HYG Database (David Nash), licensed CC BY-SA 4.0. The project's corrections and the map and orrery data built on it are derivatives, so share-alike applies to derived star data. Gaia DR3 and the ATNF Pulsar Catalogue ask for acknowledgement, given on the licensing page. Wiki text is CC BY-SA 4.0. Articles are copyright the Pax Abyssi team. Each image carries its own licence line.
Press / anywhere on the site to search once search opens; it will cover the wiki, articles, the forum and star names.